git.oblomov.eu Git - git/blob - commit-graph.c

   1 #include "git-compat-util.h"
   2 #include "config.h"
   3 #include "lockfile.h"
   4 #include "pack.h"
   5 #include "packfile.h"
   6 #include "commit.h"
   7 #include "object.h"
   8 #include "refs.h"
   9 #include "revision.h"
  10 #include "sha1-lookup.h"
  11 #include "commit-graph.h"
  12 #include "object-store.h"
  13 #include "alloc.h"
  14 #include "hashmap.h"
  15 #include "replace-object.h"
  16 #include "progress.h"
  17 #include "bloom.h"
  18 #include "commit-slab.h"
  19 #include "shallow.h"
  20 #include "json-writer.h"
  21 #include "trace2.h"
  22
  23 void git_test_write_commit_graph_or_die(void)
  24 {
  25         int flags = 0;
  26         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0))
  27                 return;
  28
  29         if (git_env_bool(GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS, 0))
  30                 flags = COMMIT_GRAPH_WRITE_BLOOM_FILTERS;
  31
  32         if (write_commit_graph_reachable(the_repository->objects->odb,
  33                                          flags, NULL))
  34                 die("failed to write commit-graph under GIT_TEST_COMMIT_GRAPH");
  35 }
  36
  37 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  38 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  39 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  40 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  41 #define GRAPH_CHUNKID_EXTRAEDGES 0x45444745 /* "EDGE" */
  42 #define GRAPH_CHUNKID_BLOOMINDEXES 0x42494458 /* "BIDX" */
  43 #define GRAPH_CHUNKID_BLOOMDATA 0x42444154 /* "BDAT" */
  44 #define GRAPH_CHUNKID_BASE 0x42415345 /* "BASE" */
  45 #define MAX_NUM_CHUNKS 7
  46
  47 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
  48
  49 #define GRAPH_VERSION_1 0x1
  50 #define GRAPH_VERSION GRAPH_VERSION_1
  51
  52 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
  53 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  54 #define GRAPH_PARENT_NONE 0x70000000
  55
  56 #define GRAPH_LAST_EDGE 0x80000000
  57
  58 #define GRAPH_HEADER_SIZE 8
  59 #define GRAPH_FANOUT_SIZE (4 * 256)
  60 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  61 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  62                         + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
  63
  64 /* Remember to update object flag allocation in object.h */
  65 #define REACHABLE       (1u<<15)
  66
  67 /* Keep track of the order in which commits are added to our list. */
  68 define_commit_slab(commit_pos, int);
  69 static struct commit_pos commit_pos = COMMIT_SLAB_INIT(1, commit_pos);
  70
  71 static void set_commit_pos(struct repository *r, const struct object_id *oid)
  72 {
  73         static int32_t max_pos;
  74         struct commit *commit = lookup_commit(r, oid);
  75
  76         if (!commit)
  77                 return; /* should never happen, but be lenient */
  78
  79         *commit_pos_at(&commit_pos, commit) = max_pos++;
  80 }
  81
  82 static int commit_pos_cmp(const void *va, const void *vb)
  83 {
  84         const struct commit *a = *(const struct commit **)va;
  85         const struct commit *b = *(const struct commit **)vb;
  86         return commit_pos_at(&commit_pos, a) -
  87                commit_pos_at(&commit_pos, b);
  88 }
  89
  90 define_commit_slab(commit_graph_data_slab, struct commit_graph_data);
  91 static struct commit_graph_data_slab commit_graph_data_slab =
  92         COMMIT_SLAB_INIT(1, commit_graph_data_slab);
  93
  94 uint32_t commit_graph_position(const struct commit *c)
  95 {
  96         struct commit_graph_data *data =
  97                 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
  98
  99         return data ? data->graph_pos : COMMIT_NOT_FROM_GRAPH;
 100 }
 101
 102 uint32_t commit_graph_generation(const struct commit *c)
 103 {
 104         struct commit_graph_data *data =
 105                 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
 106
 107         if (!data)
 108                 return GENERATION_NUMBER_INFINITY;
 109         else if (data->graph_pos == COMMIT_NOT_FROM_GRAPH)
 110                 return GENERATION_NUMBER_INFINITY;
 111
 112         return data->generation;
 113 }
 114
 115 static struct commit_graph_data *commit_graph_data_at(const struct commit *c)
 116 {
 117         unsigned int i, nth_slab;
 118         struct commit_graph_data *data =
 119                 commit_graph_data_slab_peek(&commit_graph_data_slab, c);
 120
 121         if (data)
 122                 return data;
 123
 124         nth_slab = c->index / commit_graph_data_slab.slab_size;
 125         data = commit_graph_data_slab_at(&commit_graph_data_slab, c);
 126
 127         /*
 128          * commit-slab initializes elements with zero, overwrite this with
 129          * COMMIT_NOT_FROM_GRAPH for graph_pos.
 130          *
 131          * We avoid initializing generation with checking if graph position
 132          * is not COMMIT_NOT_FROM_GRAPH.
 133          */
 134         for (i = 0; i < commit_graph_data_slab.slab_size; i++) {
 135                 commit_graph_data_slab.slab[nth_slab][i].graph_pos =
 136                         COMMIT_NOT_FROM_GRAPH;
 137         }
 138
 139         return data;
 140 }
 141
 142 static int commit_gen_cmp(const void *va, const void *vb)
 143 {
 144         const struct commit *a = *(const struct commit **)va;
 145         const struct commit *b = *(const struct commit **)vb;
 146
 147         uint32_t generation_a = commit_graph_generation(a);
 148         uint32_t generation_b = commit_graph_generation(b);
 149         /* lower generation commits first */
 150         if (generation_a < generation_b)
 151                 return -1;
 152         else if (generation_a > generation_b)
 153                 return 1;
 154
 155         /* use date as a heuristic when generations are equal */
 156         if (a->date < b->date)
 157                 return -1;
 158         else if (a->date > b->date)
 159                 return 1;
 160         return 0;
 161 }
 162
 163 char *get_commit_graph_filename(struct object_directory *obj_dir)
 164 {
 165         return xstrfmt("%s/info/commit-graph", obj_dir->path);
 166 }
 167
 168 static char *get_split_graph_filename(struct object_directory *odb,
 169                                       const char *oid_hex)
 170 {
 171         return xstrfmt("%s/info/commit-graphs/graph-%s.graph", odb->path,
 172                        oid_hex);
 173 }
 174
 175 static char *get_chain_filename(struct object_directory *odb)
 176 {
 177         return xstrfmt("%s/info/commit-graphs/commit-graph-chain", odb->path);
 178 }
 179
 180 static uint8_t oid_version(void)
 181 {
 182         return 1;
 183 }
 184
 185 static struct commit_graph *alloc_commit_graph(void)
 186 {
 187         struct commit_graph *g = xcalloc(1, sizeof(*g));
 188
 189         return g;
 190 }
 191
 192 extern int read_replace_refs;
 193
 194 static int commit_graph_compatible(struct repository *r)
 195 {
 196         if (!r->gitdir)
 197                 return 0;
 198
 199         if (read_replace_refs) {
 200                 prepare_replace_object(r);
 201                 if (hashmap_get_size(&r->objects->replace_map->map))
 202                         return 0;
 203         }
 204
 205         prepare_commit_graft(r);
 206         if (r->parsed_objects &&
 207             (r->parsed_objects->grafts_nr || r->parsed_objects->substituted_parent))
 208                 return 0;
 209         if (is_repository_shallow(r))
 210                 return 0;
 211
 212         return 1;
 213 }
 214
 215 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
 216 {
 217         *fd = git_open(graph_file);
 218         if (*fd < 0)
 219                 return 0;
 220         if (fstat(*fd, st)) {
 221                 close(*fd);
 222                 return 0;
 223         }
 224         return 1;
 225 }
 226
 227 struct commit_graph *load_commit_graph_one_fd_st(struct repository *r,
 228                                                  int fd, struct stat *st,
 229                                                  struct object_directory *odb)
 230 {
 231         void *graph_map;
 232         size_t graph_size;
 233         struct commit_graph *ret;
 234
 235         graph_size = xsize_t(st->st_size);
 236
 237         if (graph_size < GRAPH_MIN_SIZE) {
 238                 close(fd);
 239                 error(_("commit-graph file is too small"));
 240                 return NULL;
 241         }
 242         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
 243         close(fd);
 244         ret = parse_commit_graph(r, graph_map, graph_size);
 245
 246         if (ret)
 247                 ret->odb = odb;
 248         else
 249                 munmap(graph_map, graph_size);
 250
 251         return ret;
 252 }
 253
 254 static int verify_commit_graph_lite(struct commit_graph *g)
 255 {
 256         /*
 257          * Basic validation shared between parse_commit_graph()
 258          * which'll be called every time the graph is used, and the
 259          * much more expensive verify_commit_graph() used by
 260          * "commit-graph verify".
 261          *
 262          * There should only be very basic checks here to ensure that
 263          * we don't e.g. segfault in fill_commit_in_graph(), but
 264          * because this is a very hot codepath nothing that e.g. loops
 265          * over g->num_commits, or runs a checksum on the commit-graph
 266          * itself.
 267          */
 268         if (!g->chunk_oid_fanout) {
 269                 error("commit-graph is missing the OID Fanout chunk");
 270                 return 1;
 271         }
 272         if (!g->chunk_oid_lookup) {
 273                 error("commit-graph is missing the OID Lookup chunk");
 274                 return 1;
 275         }
 276         if (!g->chunk_commit_data) {
 277                 error("commit-graph is missing the Commit Data chunk");
 278                 return 1;
 279         }
 280
 281         return 0;
 282 }
 283
 284 struct commit_graph *parse_commit_graph(struct repository *r,
 285                                         void *graph_map, size_t graph_size)
 286 {
 287         const unsigned char *data, *chunk_lookup;
 288         uint32_t i;
 289         struct commit_graph *graph;
 290         uint64_t next_chunk_offset;
 291         uint32_t graph_signature;
 292         unsigned char graph_version, hash_version;
 293
 294         if (!graph_map)
 295                 return NULL;
 296
 297         if (graph_size < GRAPH_MIN_SIZE)
 298                 return NULL;
 299
 300         data = (const unsigned char *)graph_map;
 301
 302         graph_signature = get_be32(data);
 303         if (graph_signature != GRAPH_SIGNATURE) {
 304                 error(_("commit-graph signature %X does not match signature %X"),
 305                       graph_signature, GRAPH_SIGNATURE);
 306                 return NULL;
 307         }
 308
 309         graph_version = *(unsigned char*)(data + 4);
 310         if (graph_version != GRAPH_VERSION) {
 311                 error(_("commit-graph version %X does not match version %X"),
 312                       graph_version, GRAPH_VERSION);
 313                 return NULL;
 314         }
 315
 316         hash_version = *(unsigned char*)(data + 5);
 317         if (hash_version != oid_version()) {
 318                 error(_("commit-graph hash version %X does not match version %X"),
 319                       hash_version, oid_version());
 320                 return NULL;
 321         }
 322
 323         prepare_repo_settings(r);
 324
 325         graph = alloc_commit_graph();
 326
 327         graph->hash_len = the_hash_algo->rawsz;
 328         graph->num_chunks = *(unsigned char*)(data + 6);
 329         graph->data = graph_map;
 330         graph->data_len = graph_size;
 331
 332         if (graph_size < GRAPH_HEADER_SIZE +
 333                          (graph->num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH +
 334                          GRAPH_FANOUT_SIZE + the_hash_algo->rawsz) {
 335                 error(_("commit-graph file is too small to hold %u chunks"),
 336                       graph->num_chunks);
 337                 free(graph);
 338                 return NULL;
 339         }
 340
 341         chunk_lookup = data + 8;
 342         next_chunk_offset = get_be64(chunk_lookup + 4);
 343         for (i = 0; i < graph->num_chunks; i++) {
 344                 uint32_t chunk_id;
 345                 uint64_t chunk_offset = next_chunk_offset;
 346                 int chunk_repeated = 0;
 347
 348                 chunk_id = get_be32(chunk_lookup + 0);
 349
 350                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 351                 next_chunk_offset = get_be64(chunk_lookup + 4);
 352
 353                 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
 354                         error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 355                               (uint32_t)chunk_offset);
 356                         goto free_and_return;
 357                 }
 358
 359                 switch (chunk_id) {
 360                 case GRAPH_CHUNKID_OIDFANOUT:
 361                         if (graph->chunk_oid_fanout)
 362                                 chunk_repeated = 1;
 363                         else
 364                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 365                         break;
 366
 367                 case GRAPH_CHUNKID_OIDLOOKUP:
 368                         if (graph->chunk_oid_lookup)
 369                                 chunk_repeated = 1;
 370                         else {
 371                                 graph->chunk_oid_lookup = data + chunk_offset;
 372                                 graph->num_commits = (next_chunk_offset - chunk_offset)
 373                                                      / graph->hash_len;
 374                         }
 375                         break;
 376
 377                 case GRAPH_CHUNKID_DATA:
 378                         if (graph->chunk_commit_data)
 379                                 chunk_repeated = 1;
 380                         else
 381                                 graph->chunk_commit_data = data + chunk_offset;
 382                         break;
 383
 384                 case GRAPH_CHUNKID_EXTRAEDGES:
 385                         if (graph->chunk_extra_edges)
 386                                 chunk_repeated = 1;
 387                         else
 388                                 graph->chunk_extra_edges = data + chunk_offset;
 389                         break;
 390
 391                 case GRAPH_CHUNKID_BASE:
 392                         if (graph->chunk_base_graphs)
 393                                 chunk_repeated = 1;
 394                         else
 395                                 graph->chunk_base_graphs = data + chunk_offset;
 396                         break;
 397
 398                 case GRAPH_CHUNKID_BLOOMINDEXES:
 399                         if (graph->chunk_bloom_indexes)
 400                                 chunk_repeated = 1;
 401                         else if (r->settings.commit_graph_read_changed_paths)
 402                                 graph->chunk_bloom_indexes = data + chunk_offset;
 403                         break;
 404
 405                 case GRAPH_CHUNKID_BLOOMDATA:
 406                         if (graph->chunk_bloom_data)
 407                                 chunk_repeated = 1;
 408                         else if (r->settings.commit_graph_read_changed_paths) {
 409                                 uint32_t hash_version;
 410                                 graph->chunk_bloom_data = data + chunk_offset;
 411                                 hash_version = get_be32(data + chunk_offset);
 412
 413                                 if (hash_version != 1)
 414                                         break;
 415
 416                                 graph->bloom_filter_settings = xmalloc(sizeof(struct bloom_filter_settings));
 417                                 graph->bloom_filter_settings->hash_version = hash_version;
 418                                 graph->bloom_filter_settings->num_hashes = get_be32(data + chunk_offset + 4);
 419                                 graph->bloom_filter_settings->bits_per_entry = get_be32(data + chunk_offset + 8);
 420                                 graph->bloom_filter_settings->max_changed_paths = DEFAULT_BLOOM_MAX_CHANGES;
 421                         }
 422                         break;
 423                 }
 424
 425                 if (chunk_repeated) {
 426                         error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
 427                         goto free_and_return;
 428                 }
 429         }
 430
 431         if (graph->chunk_bloom_indexes && graph->chunk_bloom_data) {
 432                 init_bloom_filters();
 433         } else {
 434                 /* We need both the bloom chunks to exist together. Else ignore the data */
 435                 graph->chunk_bloom_indexes = NULL;
 436                 graph->chunk_bloom_data = NULL;
 437                 FREE_AND_NULL(graph->bloom_filter_settings);
 438         }
 439
 440         hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
 441
 442         if (verify_commit_graph_lite(graph))
 443                 goto free_and_return;
 444
 445         return graph;
 446
 447 free_and_return:
 448         free(graph->bloom_filter_settings);
 449         free(graph);
 450         return NULL;
 451 }
 452
 453 static struct commit_graph *load_commit_graph_one(struct repository *r,
 454                                                   const char *graph_file,
 455                                                   struct object_directory *odb)
 456 {
 457
 458         struct stat st;
 459         int fd;
 460         struct commit_graph *g;
 461         int open_ok = open_commit_graph(graph_file, &fd, &st);
 462
 463         if (!open_ok)
 464                 return NULL;
 465
 466         g = load_commit_graph_one_fd_st(r, fd, &st, odb);
 467
 468         if (g)
 469                 g->filename = xstrdup(graph_file);
 470
 471         return g;
 472 }
 473
 474 static struct commit_graph *load_commit_graph_v1(struct repository *r,
 475                                                  struct object_directory *odb)
 476 {
 477         char *graph_name = get_commit_graph_filename(odb);
 478         struct commit_graph *g = load_commit_graph_one(r, graph_name, odb);
 479         free(graph_name);
 480
 481         return g;
 482 }
 483
 484 static int add_graph_to_chain(struct commit_graph *g,
 485                               struct commit_graph *chain,
 486                               struct object_id *oids,
 487                               int n)
 488 {
 489         struct commit_graph *cur_g = chain;
 490
 491         if (n && !g->chunk_base_graphs) {
 492                 warning(_("commit-graph has no base graphs chunk"));
 493                 return 0;
 494         }
 495
 496         while (n) {
 497                 n--;
 498
 499                 if (!cur_g ||
 500                     !oideq(&oids[n], &cur_g->oid) ||
 501                     !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
 502                         warning(_("commit-graph chain does not match"));
 503                         return 0;
 504                 }
 505
 506                 cur_g = cur_g->base_graph;
 507         }
 508
 509         g->base_graph = chain;
 510
 511         if (chain)
 512                 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
 513
 514         return 1;
 515 }
 516
 517 static struct commit_graph *load_commit_graph_chain(struct repository *r,
 518                                                     struct object_directory *odb)
 519 {
 520         struct commit_graph *graph_chain = NULL;
 521         struct strbuf line = STRBUF_INIT;
 522         struct stat st;
 523         struct object_id *oids;
 524         int i = 0, valid = 1, count;
 525         char *chain_name = get_chain_filename(odb);
 526         FILE *fp;
 527         int stat_res;
 528
 529         fp = fopen(chain_name, "r");
 530         stat_res = stat(chain_name, &st);
 531         free(chain_name);
 532
 533         if (!fp ||
 534             stat_res ||
 535             st.st_size <= the_hash_algo->hexsz)
 536                 return NULL;
 537
 538         count = st.st_size / (the_hash_algo->hexsz + 1);
 539         oids = xcalloc(count, sizeof(struct object_id));
 540
 541         prepare_alt_odb(r);
 542
 543         for (i = 0; i < count; i++) {
 544                 struct object_directory *odb;
 545
 546                 if (strbuf_getline_lf(&line, fp) == EOF)
 547                         break;
 548
 549                 if (get_oid_hex(line.buf, &oids[i])) {
 550                         warning(_("invalid commit-graph chain: line '%s' not a hash"),
 551                                 line.buf);
 552                         valid = 0;
 553                         break;
 554                 }
 555
 556                 valid = 0;
 557                 for (odb = r->objects->odb; odb; odb = odb->next) {
 558                         char *graph_name = get_split_graph_filename(odb, line.buf);
 559                         struct commit_graph *g = load_commit_graph_one(r, graph_name, odb);
 560
 561                         free(graph_name);
 562
 563                         if (g) {
 564                                 if (add_graph_to_chain(g, graph_chain, oids, i)) {
 565                                         graph_chain = g;
 566                                         valid = 1;
 567                                 }
 568
 569                                 break;
 570                         }
 571                 }
 572
 573                 if (!valid) {
 574                         warning(_("unable to find all commit-graph files"));
 575                         break;
 576                 }
 577         }
 578
 579         free(oids);
 580         fclose(fp);
 581         strbuf_release(&line);
 582
 583         return graph_chain;
 584 }
 585
 586 struct commit_graph *read_commit_graph_one(struct repository *r,
 587                                            struct object_directory *odb)
 588 {
 589         struct commit_graph *g = load_commit_graph_v1(r, odb);
 590
 591         if (!g)
 592                 g = load_commit_graph_chain(r, odb);
 593
 594         return g;
 595 }
 596
 597 static void prepare_commit_graph_one(struct repository *r,
 598                                      struct object_directory *odb)
 599 {
 600
 601         if (r->objects->commit_graph)
 602                 return;
 603
 604         r->objects->commit_graph = read_commit_graph_one(r, odb);
 605 }
 606
 607 /*
 608  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 609  *
 610  * On the first invocation, this function attempts to load the commit
 611  * graph if the_repository is configured to have one.
 612  */
 613 static int prepare_commit_graph(struct repository *r)
 614 {
 615         struct object_directory *odb;
 616
 617         /*
 618          * This must come before the "already attempted?" check below, because
 619          * we want to disable even an already-loaded graph file.
 620          */
 621         if (r->commit_graph_disabled)
 622                 return 0;
 623
 624         if (r->objects->commit_graph_attempted)
 625                 return !!r->objects->commit_graph;
 626         r->objects->commit_graph_attempted = 1;
 627
 628         prepare_repo_settings(r);
 629
 630         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 631             r->settings.core_commit_graph != 1)
 632                 /*
 633                  * This repository is not configured to use commit graphs, so
 634                  * do not load one. (But report commit_graph_attempted anyway
 635                  * so that commit graph loading is not attempted again for this
 636                  * repository.)
 637                  */
 638                 return 0;
 639
 640         if (!commit_graph_compatible(r))
 641                 return 0;
 642
 643         prepare_alt_odb(r);
 644         for (odb = r->objects->odb;
 645              !r->objects->commit_graph && odb;
 646              odb = odb->next)
 647                 prepare_commit_graph_one(r, odb);
 648         return !!r->objects->commit_graph;
 649 }
 650
 651 int generation_numbers_enabled(struct repository *r)
 652 {
 653         uint32_t first_generation;
 654         struct commit_graph *g;
 655         if (!prepare_commit_graph(r))
 656                return 0;
 657
 658         g = r->objects->commit_graph;
 659
 660         if (!g->num_commits)
 661                 return 0;
 662
 663         first_generation = get_be32(g->chunk_commit_data +
 664                                     g->hash_len + 8) >> 2;
 665
 666         return !!first_generation;
 667 }
 668
 669 struct bloom_filter_settings *get_bloom_filter_settings(struct repository *r)
 670 {
 671         struct commit_graph *g = r->objects->commit_graph;
 672         while (g) {
 673                 if (g->bloom_filter_settings)
 674                         return g->bloom_filter_settings;
 675                 g = g->base_graph;
 676         }
 677         return NULL;
 678 }
 679
 680 static void close_commit_graph_one(struct commit_graph *g)
 681 {
 682         if (!g)
 683                 return;
 684
 685         close_commit_graph_one(g->base_graph);
 686         free_commit_graph(g);
 687 }
 688
 689 void close_commit_graph(struct raw_object_store *o)
 690 {
 691         close_commit_graph_one(o->commit_graph);
 692         o->commit_graph = NULL;
 693 }
 694
 695 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 696 {
 697         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 698                             g->chunk_oid_lookup, g->hash_len, pos);
 699 }
 700
 701 static void load_oid_from_graph(struct commit_graph *g,
 702                                 uint32_t pos,
 703                                 struct object_id *oid)
 704 {
 705         uint32_t lex_index;
 706
 707         while (g && pos < g->num_commits_in_base)
 708                 g = g->base_graph;
 709
 710         if (!g)
 711                 BUG("NULL commit-graph");
 712
 713         if (pos >= g->num_commits + g->num_commits_in_base)
 714                 die(_("invalid commit position. commit-graph is likely corrupt"));
 715
 716         lex_index = pos - g->num_commits_in_base;
 717
 718         hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
 719 }
 720
 721 static struct commit_list **insert_parent_or_die(struct repository *r,
 722                                                  struct commit_graph *g,
 723                                                  uint32_t pos,
 724                                                  struct commit_list **pptr)
 725 {
 726         struct commit *c;
 727         struct object_id oid;
 728
 729         if (pos >= g->num_commits + g->num_commits_in_base)
 730                 die("invalid parent position %"PRIu32, pos);
 731
 732         load_oid_from_graph(g, pos, &oid);
 733         c = lookup_commit(r, &oid);
 734         if (!c)
 735                 die(_("could not find commit %s"), oid_to_hex(&oid));
 736         commit_graph_data_at(c)->graph_pos = pos;
 737         return &commit_list_insert(c, pptr)->next;
 738 }
 739
 740 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 741 {
 742         const unsigned char *commit_data;
 743         struct commit_graph_data *graph_data;
 744         uint32_t lex_index;
 745
 746         while (pos < g->num_commits_in_base)
 747                 g = g->base_graph;
 748
 749         lex_index = pos - g->num_commits_in_base;
 750         commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
 751
 752         graph_data = commit_graph_data_at(item);
 753         graph_data->graph_pos = pos;
 754         graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 755 }
 756
 757 static inline void set_commit_tree(struct commit *c, struct tree *t)
 758 {
 759         c->maybe_tree = t;
 760 }
 761
 762 static int fill_commit_in_graph(struct repository *r,
 763                                 struct commit *item,
 764                                 struct commit_graph *g, uint32_t pos)
 765 {
 766         uint32_t edge_value;
 767         uint32_t *parent_data_ptr;
 768         uint64_t date_low, date_high;
 769         struct commit_list **pptr;
 770         struct commit_graph_data *graph_data;
 771         const unsigned char *commit_data;
 772         uint32_t lex_index;
 773
 774         while (pos < g->num_commits_in_base)
 775                 g = g->base_graph;
 776
 777         if (pos >= g->num_commits + g->num_commits_in_base)
 778                 die(_("invalid commit position. commit-graph is likely corrupt"));
 779
 780         /*
 781          * Store the "full" position, but then use the
 782          * "local" position for the rest of the calculation.
 783          */
 784         graph_data = commit_graph_data_at(item);
 785         graph_data->graph_pos = pos;
 786         lex_index = pos - g->num_commits_in_base;
 787
 788         commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
 789
 790         item->object.parsed = 1;
 791
 792         set_commit_tree(item, NULL);
 793
 794         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 795         date_low = get_be32(commit_data + g->hash_len + 12);
 796         item->date = (timestamp_t)((date_high << 32) | date_low);
 797
 798         graph_data->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 799
 800         pptr = &item->parents;
 801
 802         edge_value = get_be32(commit_data + g->hash_len);
 803         if (edge_value == GRAPH_PARENT_NONE)
 804                 return 1;
 805         pptr = insert_parent_or_die(r, g, edge_value, pptr);
 806
 807         edge_value = get_be32(commit_data + g->hash_len + 4);
 808         if (edge_value == GRAPH_PARENT_NONE)
 809                 return 1;
 810         if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
 811                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
 812                 return 1;
 813         }
 814
 815         parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
 816                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 817         do {
 818                 edge_value = get_be32(parent_data_ptr);
 819                 pptr = insert_parent_or_die(r, g,
 820                                             edge_value & GRAPH_EDGE_LAST_MASK,
 821                                             pptr);
 822                 parent_data_ptr++;
 823         } while (!(edge_value & GRAPH_LAST_EDGE));
 824
 825         return 1;
 826 }
 827
 828 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 829 {
 830         uint32_t graph_pos = commit_graph_position(item);
 831         if (graph_pos != COMMIT_NOT_FROM_GRAPH) {
 832                 *pos = graph_pos;
 833                 return 1;
 834         } else {
 835                 struct commit_graph *cur_g = g;
 836                 uint32_t lex_index;
 837
 838                 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
 839                         cur_g = cur_g->base_graph;
 840
 841                 if (cur_g) {
 842                         *pos = lex_index + cur_g->num_commits_in_base;
 843                         return 1;
 844                 }
 845
 846                 return 0;
 847         }
 848 }
 849
 850 static int parse_commit_in_graph_one(struct repository *r,
 851                                      struct commit_graph *g,
 852                                      struct commit *item)
 853 {
 854         uint32_t pos;
 855
 856         if (item->object.parsed)
 857                 return 1;
 858
 859         if (find_commit_in_graph(item, g, &pos))
 860                 return fill_commit_in_graph(r, item, g, pos);
 861
 862         return 0;
 863 }
 864
 865 int parse_commit_in_graph(struct repository *r, struct commit *item)
 866 {
 867         static int checked_env = 0;
 868
 869         if (!checked_env &&
 870             git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE, 0))
 871                 die("dying as requested by the '%s' variable on commit-graph parse!",
 872                     GIT_TEST_COMMIT_GRAPH_DIE_ON_PARSE);
 873         checked_env = 1;
 874
 875         if (!prepare_commit_graph(r))
 876                 return 0;
 877         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
 878 }
 879
 880 void load_commit_graph_info(struct repository *r, struct commit *item)
 881 {
 882         uint32_t pos;
 883         if (!prepare_commit_graph(r))
 884                 return;
 885         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 886                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 887 }
 888
 889 static struct tree *load_tree_for_commit(struct repository *r,
 890                                          struct commit_graph *g,
 891                                          struct commit *c)
 892 {
 893         struct object_id oid;
 894         const unsigned char *commit_data;
 895         uint32_t graph_pos = commit_graph_position(c);
 896
 897         while (graph_pos < g->num_commits_in_base)
 898                 g = g->base_graph;
 899
 900         commit_data = g->chunk_commit_data +
 901                         GRAPH_DATA_WIDTH * (graph_pos - g->num_commits_in_base);
 902
 903         hashcpy(oid.hash, commit_data);
 904         set_commit_tree(c, lookup_tree(r, &oid));
 905
 906         return c->maybe_tree;
 907 }
 908
 909 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
 910                                                  struct commit_graph *g,
 911                                                  const struct commit *c)
 912 {
 913         if (c->maybe_tree)
 914                 return c->maybe_tree;
 915         if (commit_graph_position(c) == COMMIT_NOT_FROM_GRAPH)
 916                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 917
 918         return load_tree_for_commit(r, g, (struct commit *)c);
 919 }
 920
 921 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 922 {
 923         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
 924 }
 925
 926 struct packed_commit_list {
 927         struct commit **list;
 928         int nr;
 929         int alloc;
 930 };
 931
 932 struct packed_oid_list {
 933         struct object_id *list;
 934         int nr;
 935         int alloc;
 936 };
 937
 938 struct write_commit_graph_context {
 939         struct repository *r;
 940         struct object_directory *odb;
 941         char *graph_name;
 942         struct packed_oid_list oids;
 943         struct packed_commit_list commits;
 944         int num_extra_edges;
 945         unsigned long approx_nr_objects;
 946         struct progress *progress;
 947         int progress_done;
 948         uint64_t progress_cnt;
 949
 950         char *base_graph_name;
 951         int num_commit_graphs_before;
 952         int num_commit_graphs_after;
 953         char **commit_graph_filenames_before;
 954         char **commit_graph_filenames_after;
 955         char **commit_graph_hash_after;
 956         uint32_t new_num_commits_in_base;
 957         struct commit_graph *new_base_graph;
 958
 959         unsigned append:1,
 960                  report_progress:1,
 961                  split:1,
 962                  changed_paths:1,
 963                  order_by_pack:1;
 964
 965         const struct split_commit_graph_opts *split_opts;
 966         size_t total_bloom_filter_data_size;
 967         const struct bloom_filter_settings *bloom_settings;
 968
 969         int count_bloom_filter_computed;
 970         int count_bloom_filter_not_computed;
 971         int count_bloom_filter_trunc_large;
 972 };
 973
 974 static int write_graph_chunk_fanout(struct hashfile *f,
 975                                     struct write_commit_graph_context *ctx)
 976 {
 977         int i, count = 0;
 978         struct commit **list = ctx->commits.list;
 979
 980         /*
 981          * Write the first-level table (the list is sorted,
 982          * but we use a 256-entry lookup to be able to avoid
 983          * having to do eight extra binary search iterations).
 984          */
 985         for (i = 0; i < 256; i++) {
 986                 while (count < ctx->commits.nr) {
 987                         if ((*list)->object.oid.hash[0] != i)
 988                                 break;
 989                         display_progress(ctx->progress, ++ctx->progress_cnt);
 990                         count++;
 991                         list++;
 992                 }
 993
 994                 hashwrite_be32(f, count);
 995         }
 996
 997         return 0;
 998 }
 999
1000 static int write_graph_chunk_oids(struct hashfile *f,
1001                                   struct write_commit_graph_context *ctx)
1002 {
1003         struct commit **list = ctx->commits.list;
1004         int count;
1005         for (count = 0; count < ctx->commits.nr; count++, list++) {
1006                 display_progress(ctx->progress, ++ctx->progress_cnt);
1007                 hashwrite(f, (*list)->object.oid.hash, the_hash_algo->rawsz);
1008         }
1009
1010         return 0;
1011 }
1012
1013 static const unsigned char *commit_to_sha1(size_t index, void *table)
1014 {
1015         struct commit **commits = table;
1016         return commits[index]->object.oid.hash;
1017 }
1018
1019 static int write_graph_chunk_data(struct hashfile *f,
1020                                   struct write_commit_graph_context *ctx)
1021 {
1022         struct commit **list = ctx->commits.list;
1023         struct commit **last = ctx->commits.list + ctx->commits.nr;
1024         uint32_t num_extra_edges = 0;
1025
1026         while (list < last) {
1027                 struct commit_list *parent;
1028                 struct object_id *tree;
1029                 int edge_value;
1030                 uint32_t packedDate[2];
1031                 display_progress(ctx->progress, ++ctx->progress_cnt);
1032
1033                 if (parse_commit_no_graph(*list))
1034                         die(_("unable to parse commit %s"),
1035                                 oid_to_hex(&(*list)->object.oid));
1036                 tree = get_commit_tree_oid(*list);
1037                 hashwrite(f, tree->hash, the_hash_algo->rawsz);
1038
1039                 parent = (*list)->parents;
1040
1041                 if (!parent)
1042                         edge_value = GRAPH_PARENT_NONE;
1043                 else {
1044                         edge_value = sha1_pos(parent->item->object.oid.hash,
1045                                               ctx->commits.list,
1046                                               ctx->commits.nr,
1047                                               commit_to_sha1);
1048
1049                         if (edge_value >= 0)
1050                                 edge_value += ctx->new_num_commits_in_base;
1051                         else if (ctx->new_base_graph) {
1052                                 uint32_t pos;
1053                                 if (find_commit_in_graph(parent->item,
1054                                                          ctx->new_base_graph,
1055                                                          &pos))
1056                                         edge_value = pos;
1057                         }
1058
1059                         if (edge_value < 0)
1060                                 BUG("missing parent %s for commit %s",
1061                                     oid_to_hex(&parent->item->object.oid),
1062                                     oid_to_hex(&(*list)->object.oid));
1063                 }
1064
1065                 hashwrite_be32(f, edge_value);
1066
1067                 if (parent)
1068                         parent = parent->next;
1069
1070                 if (!parent)
1071                         edge_value = GRAPH_PARENT_NONE;
1072                 else if (parent->next)
1073                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
1074                 else {
1075                         edge_value = sha1_pos(parent->item->object.oid.hash,
1076                                               ctx->commits.list,
1077                                               ctx->commits.nr,
1078                                               commit_to_sha1);
1079
1080                         if (edge_value >= 0)
1081                                 edge_value += ctx->new_num_commits_in_base;
1082                         else if (ctx->new_base_graph) {
1083                                 uint32_t pos;
1084                                 if (find_commit_in_graph(parent->item,
1085                                                          ctx->new_base_graph,
1086                                                          &pos))
1087                                         edge_value = pos;
1088                         }
1089
1090                         if (edge_value < 0)
1091                                 BUG("missing parent %s for commit %s",
1092                                     oid_to_hex(&parent->item->object.oid),
1093                                     oid_to_hex(&(*list)->object.oid));
1094                 }
1095
1096                 hashwrite_be32(f, edge_value);
1097
1098                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1099                         do {
1100                                 num_extra_edges++;
1101                                 parent = parent->next;
1102                         } while (parent);
1103                 }
1104
1105                 if (sizeof((*list)->date) > 4)
1106                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1107                 else
1108                         packedDate[0] = 0;
1109
1110                 packedDate[0] |= htonl(commit_graph_data_at(*list)->generation << 2);
1111
1112                 packedDate[1] = htonl((*list)->date);
1113                 hashwrite(f, packedDate, 8);
1114
1115                 list++;
1116         }
1117
1118         return 0;
1119 }
1120
1121 static int write_graph_chunk_extra_edges(struct hashfile *f,
1122                                          struct write_commit_graph_context *ctx)
1123 {
1124         struct commit **list = ctx->commits.list;
1125         struct commit **last = ctx->commits.list + ctx->commits.nr;
1126         struct commit_list *parent;
1127
1128         while (list < last) {
1129                 int num_parents = 0;
1130
1131                 display_progress(ctx->progress, ++ctx->progress_cnt);
1132
1133                 for (parent = (*list)->parents; num_parents < 3 && parent;
1134                      parent = parent->next)
1135                         num_parents++;
1136
1137                 if (num_parents <= 2) {
1138                         list++;
1139                         continue;
1140                 }
1141
1142                 /* Since num_parents > 2, this initializer is safe. */
1143                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1144                         int edge_value = sha1_pos(parent->item->object.oid.hash,
1145                                                   ctx->commits.list,
1146                                                   ctx->commits.nr,
1147                                                   commit_to_sha1);
1148
1149                         if (edge_value >= 0)
1150                                 edge_value += ctx->new_num_commits_in_base;
1151                         else if (ctx->new_base_graph) {
1152                                 uint32_t pos;
1153                                 if (find_commit_in_graph(parent->item,
1154                                                          ctx->new_base_graph,
1155                                                          &pos))
1156                                         edge_value = pos;
1157                         }
1158
1159                         if (edge_value < 0)
1160                                 BUG("missing parent %s for commit %s",
1161                                     oid_to_hex(&parent->item->object.oid),
1162                                     oid_to_hex(&(*list)->object.oid));
1163                         else if (!parent->next)
1164                                 edge_value |= GRAPH_LAST_EDGE;
1165
1166                         hashwrite_be32(f, edge_value);
1167                 }
1168
1169                 list++;
1170         }
1171
1172         return 0;
1173 }
1174
1175 static int write_graph_chunk_bloom_indexes(struct hashfile *f,
1176                                            struct write_commit_graph_context *ctx)
1177 {
1178         struct commit **list = ctx->commits.list;
1179         struct commit **last = ctx->commits.list + ctx->commits.nr;
1180         uint32_t cur_pos = 0;
1181
1182         while (list < last) {
1183                 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list);
1184                 size_t len = filter ? filter->len : 0;
1185                 cur_pos += len;
1186                 display_progress(ctx->progress, ++ctx->progress_cnt);
1187                 hashwrite_be32(f, cur_pos);
1188                 list++;
1189         }
1190
1191         return 0;
1192 }
1193
1194 static void trace2_bloom_filter_settings(struct write_commit_graph_context *ctx)
1195 {
1196         struct json_writer jw = JSON_WRITER_INIT;
1197
1198         jw_object_begin(&jw, 0);
1199         jw_object_intmax(&jw, "hash_version", ctx->bloom_settings->hash_version);
1200         jw_object_intmax(&jw, "num_hashes", ctx->bloom_settings->num_hashes);
1201         jw_object_intmax(&jw, "bits_per_entry", ctx->bloom_settings->bits_per_entry);
1202         jw_object_intmax(&jw, "max_changed_paths", ctx->bloom_settings->max_changed_paths);
1203         jw_end(&jw);
1204
1205         trace2_data_json("bloom", ctx->r, "settings", &jw);
1206
1207         jw_release(&jw);
1208 }
1209
1210 static int write_graph_chunk_bloom_data(struct hashfile *f,
1211                                         struct write_commit_graph_context *ctx)
1212 {
1213         struct commit **list = ctx->commits.list;
1214         struct commit **last = ctx->commits.list + ctx->commits.nr;
1215
1216         trace2_bloom_filter_settings(ctx);
1217
1218         hashwrite_be32(f, ctx->bloom_settings->hash_version);
1219         hashwrite_be32(f, ctx->bloom_settings->num_hashes);
1220         hashwrite_be32(f, ctx->bloom_settings->bits_per_entry);
1221
1222         while (list < last) {
1223                 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list);
1224                 size_t len = filter ? filter->len : 0;
1225
1226                 display_progress(ctx->progress, ++ctx->progress_cnt);
1227                 if (len)
1228                         hashwrite(f, filter->data, len * sizeof(unsigned char));
1229                 list++;
1230         }
1231
1232         return 0;
1233 }
1234
1235 static int oid_compare(const void *_a, const void *_b)
1236 {
1237         const struct object_id *a = (const struct object_id *)_a;
1238         const struct object_id *b = (const struct object_id *)_b;
1239         return oidcmp(a, b);
1240 }
1241
1242 static int add_packed_commits(const struct object_id *oid,
1243                               struct packed_git *pack,
1244                               uint32_t pos,
1245                               void *data)
1246 {
1247         struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1248         enum object_type type;
1249         off_t offset = nth_packed_object_offset(pack, pos);
1250         struct object_info oi = OBJECT_INFO_INIT;
1251
1252         if (ctx->progress)
1253                 display_progress(ctx->progress, ++ctx->progress_done);
1254
1255         oi.typep = &type;
1256         if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1257                 die(_("unable to get type of object %s"), oid_to_hex(oid));
1258
1259         if (type != OBJ_COMMIT)
1260                 return 0;
1261
1262         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1263         oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1264         ctx->oids.nr++;
1265
1266         set_commit_pos(ctx->r, oid);
1267
1268         return 0;
1269 }
1270
1271 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1272 {
1273         struct commit_list *parent;
1274         for (parent = commit->parents; parent; parent = parent->next) {
1275                 if (!(parent->item->object.flags & REACHABLE)) {
1276                         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1277                         oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1278                         ctx->oids.nr++;
1279                         parent->item->object.flags |= REACHABLE;
1280                 }
1281         }
1282 }
1283
1284 static void close_reachable(struct write_commit_graph_context *ctx)
1285 {
1286         int i;
1287         struct commit *commit;
1288         enum commit_graph_split_flags flags = ctx->split_opts ?
1289                 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1290
1291         if (ctx->report_progress)
1292                 ctx->progress = start_delayed_progress(
1293                                         _("Loading known commits in commit graph"),
1294                                         ctx->oids.nr);
1295         for (i = 0; i < ctx->oids.nr; i++) {
1296                 display_progress(ctx->progress, i + 1);
1297                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1298                 if (commit)
1299                         commit->object.flags |= REACHABLE;
1300         }
1301         stop_progress(&ctx->progress);
1302
1303         /*
1304          * As this loop runs, ctx->oids.nr may grow, but not more
1305          * than the number of missing commits in the reachable
1306          * closure.
1307          */
1308         if (ctx->report_progress)
1309                 ctx->progress = start_delayed_progress(
1310                                         _("Expanding reachable commits in commit graph"),
1311                                         0);
1312         for (i = 0; i < ctx->oids.nr; i++) {
1313                 display_progress(ctx->progress, i + 1);
1314                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1315
1316                 if (!commit)
1317                         continue;
1318                 if (ctx->split) {
1319                         if ((!parse_commit(commit) &&
1320                              commit_graph_position(commit) == COMMIT_NOT_FROM_GRAPH) ||
1321                             flags == COMMIT_GRAPH_SPLIT_REPLACE)
1322                                 add_missing_parents(ctx, commit);
1323                 } else if (!parse_commit_no_graph(commit))
1324                         add_missing_parents(ctx, commit);
1325         }
1326         stop_progress(&ctx->progress);
1327
1328         if (ctx->report_progress)
1329                 ctx->progress = start_delayed_progress(
1330                                         _("Clearing commit marks in commit graph"),
1331                                         ctx->oids.nr);
1332         for (i = 0; i < ctx->oids.nr; i++) {
1333                 display_progress(ctx->progress, i + 1);
1334                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1335
1336                 if (commit)
1337                         commit->object.flags &= ~REACHABLE;
1338         }
1339         stop_progress(&ctx->progress);
1340 }
1341
1342 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1343 {
1344         int i;
1345         struct commit_list *list = NULL;
1346
1347         if (ctx->report_progress)
1348                 ctx->progress = start_delayed_progress(
1349                                         _("Computing commit graph generation numbers"),
1350                                         ctx->commits.nr);
1351         for (i = 0; i < ctx->commits.nr; i++) {
1352                 uint32_t generation = commit_graph_data_at(ctx->commits.list[i])->generation;
1353
1354                 display_progress(ctx->progress, i + 1);
1355                 if (generation != GENERATION_NUMBER_INFINITY &&
1356                     generation != GENERATION_NUMBER_ZERO)
1357                         continue;
1358
1359                 commit_list_insert(ctx->commits.list[i], &list);
1360                 while (list) {
1361                         struct commit *current = list->item;
1362                         struct commit_list *parent;
1363                         int all_parents_computed = 1;
1364                         uint32_t max_generation = 0;
1365
1366                         for (parent = current->parents; parent; parent = parent->next) {
1367                                 generation = commit_graph_data_at(parent->item)->generation;
1368
1369                                 if (generation == GENERATION_NUMBER_INFINITY ||
1370                                     generation == GENERATION_NUMBER_ZERO) {
1371                                         all_parents_computed = 0;
1372                                         commit_list_insert(parent->item, &list);
1373                                         break;
1374                                 } else if (generation > max_generation) {
1375                                         max_generation = generation;
1376                                 }
1377                         }
1378
1379                         if (all_parents_computed) {
1380                                 struct commit_graph_data *data = commit_graph_data_at(current);
1381
1382                                 data->generation = max_generation + 1;
1383                                 pop_commit(&list);
1384
1385                                 if (data->generation > GENERATION_NUMBER_MAX)
1386                                         data->generation = GENERATION_NUMBER_MAX;
1387                         }
1388                 }
1389         }
1390         stop_progress(&ctx->progress);
1391 }
1392
1393 static void trace2_bloom_filter_write_statistics(struct write_commit_graph_context *ctx)
1394 {
1395         trace2_data_intmax("commit-graph", ctx->r, "filter-computed",
1396                            ctx->count_bloom_filter_computed);
1397         trace2_data_intmax("commit-graph", ctx->r, "filter-not-computed",
1398                            ctx->count_bloom_filter_not_computed);
1399         trace2_data_intmax("commit-graph", ctx->r, "filter-trunc-large",
1400                            ctx->count_bloom_filter_trunc_large);
1401 }
1402
1403 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1404 {
1405         int i;
1406         struct progress *progress = NULL;
1407         struct commit **sorted_commits;
1408
1409         init_bloom_filters();
1410
1411         if (ctx->report_progress)
1412                 progress = start_delayed_progress(
1413                         _("Computing commit changed paths Bloom filters"),
1414                         ctx->commits.nr);
1415
1416         ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1417         COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1418
1419         if (ctx->order_by_pack)
1420                 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1421         else
1422                 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1423
1424         for (i = 0; i < ctx->commits.nr; i++) {
1425                 enum bloom_filter_computed computed = 0;
1426                 struct commit *c = sorted_commits[i];
1427                 struct bloom_filter *filter = get_or_compute_bloom_filter(
1428                         ctx->r,
1429                         c,
1430                         1,
1431                         ctx->bloom_settings,
1432                         &computed);
1433                 if (computed & BLOOM_COMPUTED) {
1434                         ctx->count_bloom_filter_computed++;
1435                         if (computed & BLOOM_TRUNC_LARGE)
1436                                 ctx->count_bloom_filter_trunc_large++;
1437                 } else if (computed & BLOOM_NOT_COMPUTED)
1438                         ctx->count_bloom_filter_not_computed++;
1439                 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1440                 display_progress(progress, i + 1);
1441         }
1442
1443         if (trace2_is_enabled())
1444                 trace2_bloom_filter_write_statistics(ctx);
1445
1446         free(sorted_commits);
1447         stop_progress(&progress);
1448 }
1449
1450 struct refs_cb_data {
1451         struct oidset *commits;
1452         struct progress *progress;
1453 };
1454
1455 static int add_ref_to_set(const char *refname,
1456                           const struct object_id *oid,
1457                           int flags, void *cb_data)
1458 {
1459         struct object_id peeled;
1460         struct refs_cb_data *data = (struct refs_cb_data *)cb_data;
1461
1462         if (!peel_ref(refname, &peeled))
1463                 oid = &peeled;
1464         if (oid_object_info(the_repository, oid, NULL) == OBJ_COMMIT)
1465                 oidset_insert(data->commits, oid);
1466
1467         display_progress(data->progress, oidset_size(data->commits));
1468
1469         return 0;
1470 }
1471
1472 int write_commit_graph_reachable(struct object_directory *odb,
1473                                  enum commit_graph_write_flags flags,
1474                                  const struct split_commit_graph_opts *split_opts)
1475 {
1476         struct oidset commits = OIDSET_INIT;
1477         struct refs_cb_data data;
1478         int result;
1479
1480         memset(&data, 0, sizeof(data));
1481         data.commits = &commits;
1482         if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
1483                 data.progress = start_delayed_progress(
1484                         _("Collecting referenced commits"), 0);
1485
1486         for_each_ref(add_ref_to_set, &data);
1487
1488         stop_progress(&data.progress);
1489
1490         result = write_commit_graph(odb, NULL, &commits,
1491                                     flags, split_opts);
1492
1493         oidset_clear(&commits);
1494         return result;
1495 }
1496
1497 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1498                                 struct string_list *pack_indexes)
1499 {
1500         uint32_t i;
1501         struct strbuf progress_title = STRBUF_INIT;
1502         struct strbuf packname = STRBUF_INIT;
1503         int dirlen;
1504
1505         strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1506         dirlen = packname.len;
1507         if (ctx->report_progress) {
1508                 strbuf_addf(&progress_title,
1509                             Q_("Finding commits for commit graph in %d pack",
1510                                "Finding commits for commit graph in %d packs",
1511                                pack_indexes->nr),
1512                             pack_indexes->nr);
1513                 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1514                 ctx->progress_done = 0;
1515         }
1516         for (i = 0; i < pack_indexes->nr; i++) {
1517                 struct packed_git *p;
1518                 strbuf_setlen(&packname, dirlen);
1519                 strbuf_addstr(&packname, pack_indexes->items[i].string);
1520                 p = add_packed_git(packname.buf, packname.len, 1);
1521                 if (!p) {
1522                         error(_("error adding pack %s"), packname.buf);
1523                         return -1;
1524                 }
1525                 if (open_pack_index(p)) {
1526                         error(_("error opening index for %s"), packname.buf);
1527                         return -1;
1528                 }
1529                 for_each_object_in_pack(p, add_packed_commits, ctx,
1530                                         FOR_EACH_OBJECT_PACK_ORDER);
1531                 close_pack(p);
1532                 free(p);
1533         }
1534
1535         stop_progress(&ctx->progress);
1536         strbuf_release(&progress_title);
1537         strbuf_release(&packname);
1538
1539         return 0;
1540 }
1541
1542 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
1543                                   struct oidset *commits)
1544 {
1545         struct oidset_iter iter;
1546         struct object_id *oid;
1547
1548         if (!oidset_size(commits))
1549                 return 0;
1550
1551         oidset_iter_init(commits, &iter);
1552         while ((oid = oidset_iter_next(&iter))) {
1553                 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1554                 oidcpy(&ctx->oids.list[ctx->oids.nr], oid);
1555                 ctx->oids.nr++;
1556         }
1557
1558         return 0;
1559 }
1560
1561 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1562 {
1563         if (ctx->report_progress)
1564                 ctx->progress = start_delayed_progress(
1565                         _("Finding commits for commit graph among packed objects"),
1566                         ctx->approx_nr_objects);
1567         for_each_packed_object(add_packed_commits, ctx,
1568                                FOR_EACH_OBJECT_PACK_ORDER);
1569         if (ctx->progress_done < ctx->approx_nr_objects)
1570                 display_progress(ctx->progress, ctx->approx_nr_objects);
1571         stop_progress(&ctx->progress);
1572 }
1573
1574 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1575 {
1576         uint32_t i, count_distinct = 1;
1577
1578         if (ctx->report_progress)
1579                 ctx->progress = start_delayed_progress(
1580                         _("Counting distinct commits in commit graph"),
1581                         ctx->oids.nr);
1582         display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1583         QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1584
1585         for (i = 1; i < ctx->oids.nr; i++) {
1586                 display_progress(ctx->progress, i + 1);
1587                 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1588                         if (ctx->split) {
1589                                 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1590
1591                                 if (!c || commit_graph_position(c) != COMMIT_NOT_FROM_GRAPH)
1592                                         continue;
1593                         }
1594
1595                         count_distinct++;
1596                 }
1597         }
1598         stop_progress(&ctx->progress);
1599
1600         return count_distinct;
1601 }
1602
1603 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1604 {
1605         uint32_t i;
1606         enum commit_graph_split_flags flags = ctx->split_opts ?
1607                 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1608
1609         ctx->num_extra_edges = 0;
1610         if (ctx->report_progress)
1611                 ctx->progress = start_delayed_progress(
1612                         _("Finding extra edges in commit graph"),
1613                         ctx->oids.nr);
1614         for (i = 0; i < ctx->oids.nr; i++) {
1615                 unsigned int num_parents;
1616
1617                 display_progress(ctx->progress, i + 1);
1618                 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1619                         continue;
1620
1621                 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1622                 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1623
1624                 if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
1625                     commit_graph_position(ctx->commits.list[ctx->commits.nr]) != COMMIT_NOT_FROM_GRAPH)
1626                         continue;
1627
1628                 if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
1629                         parse_commit(ctx->commits.list[ctx->commits.nr]);
1630                 else
1631                         parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1632
1633                 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1634                 if (num_parents > 2)
1635                         ctx->num_extra_edges += num_parents - 1;
1636
1637                 ctx->commits.nr++;
1638         }
1639         stop_progress(&ctx->progress);
1640 }
1641
1642 static int write_graph_chunk_base_1(struct hashfile *f,
1643                                     struct commit_graph *g)
1644 {
1645         int num = 0;
1646
1647         if (!g)
1648                 return 0;
1649
1650         num = write_graph_chunk_base_1(f, g->base_graph);
1651         hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1652         return num + 1;
1653 }
1654
1655 static int write_graph_chunk_base(struct hashfile *f,
1656                                   struct write_commit_graph_context *ctx)
1657 {
1658         int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1659
1660         if (num != ctx->num_commit_graphs_after - 1) {
1661                 error(_("failed to write correct number of base graph ids"));
1662                 return -1;
1663         }
1664
1665         return 0;
1666 }
1667
1668 typedef int (*chunk_write_fn)(struct hashfile *f,
1669                               struct write_commit_graph_context *ctx);
1670
1671 struct chunk_info {
1672         uint32_t id;
1673         uint64_t size;
1674         chunk_write_fn write_fn;
1675 };
1676
1677 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1678 {
1679         uint32_t i;
1680         int fd;
1681         struct hashfile *f;
1682         struct lock_file lk = LOCK_INIT;
1683         struct chunk_info chunks[MAX_NUM_CHUNKS + 1];
1684         const unsigned hashsz = the_hash_algo->rawsz;
1685         struct strbuf progress_title = STRBUF_INIT;
1686         int num_chunks = 3;
1687         uint64_t chunk_offset;
1688         struct object_id file_hash;
1689
1690         if (ctx->split) {
1691                 struct strbuf tmp_file = STRBUF_INIT;
1692
1693                 strbuf_addf(&tmp_file,
1694                             "%s/info/commit-graphs/tmp_graph_XXXXXX",
1695                             ctx->odb->path);
1696                 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1697         } else {
1698                 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1699         }
1700
1701         if (safe_create_leading_directories(ctx->graph_name)) {
1702                 UNLEAK(ctx->graph_name);
1703                 error(_("unable to create leading directories of %s"),
1704                         ctx->graph_name);
1705                 return -1;
1706         }
1707
1708         if (ctx->split) {
1709                 char *lock_name = get_chain_filename(ctx->odb);
1710
1711                 hold_lock_file_for_update_mode(&lk, lock_name,
1712                                                LOCK_DIE_ON_ERROR, 0444);
1713
1714                 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1715                 if (fd < 0) {
1716                         error(_("unable to create temporary graph layer"));
1717                         return -1;
1718                 }
1719
1720                 if (adjust_shared_perm(ctx->graph_name)) {
1721                         error(_("unable to adjust shared permissions for '%s'"),
1722                               ctx->graph_name);
1723                         return -1;
1724                 }
1725
1726                 f = hashfd(fd, ctx->graph_name);
1727         } else {
1728                 hold_lock_file_for_update_mode(&lk, ctx->graph_name,
1729                                                LOCK_DIE_ON_ERROR, 0444);
1730                 fd = lk.tempfile->fd;
1731                 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1732         }
1733
1734         chunks[0].id = GRAPH_CHUNKID_OIDFANOUT;
1735         chunks[0].size = GRAPH_FANOUT_SIZE;
1736         chunks[0].write_fn = write_graph_chunk_fanout;
1737         chunks[1].id = GRAPH_CHUNKID_OIDLOOKUP;
1738         chunks[1].size = hashsz * ctx->commits.nr;
1739         chunks[1].write_fn = write_graph_chunk_oids;
1740         chunks[2].id = GRAPH_CHUNKID_DATA;
1741         chunks[2].size = (hashsz + 16) * ctx->commits.nr;
1742         chunks[2].write_fn = write_graph_chunk_data;
1743         if (ctx->num_extra_edges) {
1744                 chunks[num_chunks].id = GRAPH_CHUNKID_EXTRAEDGES;
1745                 chunks[num_chunks].size = 4 * ctx->num_extra_edges;
1746                 chunks[num_chunks].write_fn = write_graph_chunk_extra_edges;
1747                 num_chunks++;
1748         }
1749         if (ctx->changed_paths) {
1750                 chunks[num_chunks].id = GRAPH_CHUNKID_BLOOMINDEXES;
1751                 chunks[num_chunks].size = sizeof(uint32_t) * ctx->commits.nr;
1752                 chunks[num_chunks].write_fn = write_graph_chunk_bloom_indexes;
1753                 num_chunks++;
1754                 chunks[num_chunks].id = GRAPH_CHUNKID_BLOOMDATA;
1755                 chunks[num_chunks].size = sizeof(uint32_t) * 3
1756                                           + ctx->total_bloom_filter_data_size;
1757                 chunks[num_chunks].write_fn = write_graph_chunk_bloom_data;
1758                 num_chunks++;
1759         }
1760         if (ctx->num_commit_graphs_after > 1) {
1761                 chunks[num_chunks].id = GRAPH_CHUNKID_BASE;
1762                 chunks[num_chunks].size = hashsz * (ctx->num_commit_graphs_after - 1);
1763                 chunks[num_chunks].write_fn = write_graph_chunk_base;
1764                 num_chunks++;
1765         }
1766
1767         chunks[num_chunks].id = 0;
1768         chunks[num_chunks].size = 0;
1769
1770         hashwrite_be32(f, GRAPH_SIGNATURE);
1771
1772         hashwrite_u8(f, GRAPH_VERSION);
1773         hashwrite_u8(f, oid_version());
1774         hashwrite_u8(f, num_chunks);
1775         hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1776
1777         chunk_offset = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1778         for (i = 0; i <= num_chunks; i++) {
1779                 uint32_t chunk_write[3];
1780
1781                 chunk_write[0] = htonl(chunks[i].id);
1782                 chunk_write[1] = htonl(chunk_offset >> 32);
1783                 chunk_write[2] = htonl(chunk_offset & 0xffffffff);
1784                 hashwrite(f, chunk_write, 12);
1785
1786                 chunk_offset += chunks[i].size;
1787         }
1788
1789         if (ctx->report_progress) {
1790                 strbuf_addf(&progress_title,
1791                             Q_("Writing out commit graph in %d pass",
1792                                "Writing out commit graph in %d passes",
1793                                num_chunks),
1794                             num_chunks);
1795                 ctx->progress = start_delayed_progress(
1796                         progress_title.buf,
1797                         num_chunks * ctx->commits.nr);
1798         }
1799
1800         for (i = 0; i < num_chunks; i++) {
1801                 uint64_t start_offset = f->total + f->offset;
1802
1803                 if (chunks[i].write_fn(f, ctx))
1804                         return -1;
1805
1806                 if (f->total + f->offset != start_offset + chunks[i].size)
1807                         BUG("expected to write %"PRId64" bytes to chunk %"PRIx32", but wrote %"PRId64" instead",
1808                             chunks[i].size, chunks[i].id,
1809                             f->total + f->offset - start_offset);
1810         }
1811
1812         stop_progress(&ctx->progress);
1813         strbuf_release(&progress_title);
1814
1815         if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1816                 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1817                 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1818
1819                 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1820                 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1821                 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1822                 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1823         }
1824
1825         close_commit_graph(ctx->r->objects);
1826         finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1827
1828         if (ctx->split) {
1829                 FILE *chainf = fdopen_lock_file(&lk, "w");
1830                 char *final_graph_name;
1831                 int result;
1832
1833                 close(fd);
1834
1835                 if (!chainf) {
1836                         error(_("unable to open commit-graph chain file"));
1837                         return -1;
1838                 }
1839
1840                 if (ctx->base_graph_name) {
1841                         const char *dest;
1842                         int idx = ctx->num_commit_graphs_after - 1;
1843                         if (ctx->num_commit_graphs_after > 1)
1844                                 idx--;
1845
1846                         dest = ctx->commit_graph_filenames_after[idx];
1847
1848                         if (strcmp(ctx->base_graph_name, dest)) {
1849                                 result = rename(ctx->base_graph_name, dest);
1850
1851                                 if (result) {
1852                                         error(_("failed to rename base commit-graph file"));
1853                                         return -1;
1854                                 }
1855                         }
1856                 } else {
1857                         char *graph_name = get_commit_graph_filename(ctx->odb);
1858                         unlink(graph_name);
1859                 }
1860
1861                 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1862                 final_graph_name = get_split_graph_filename(ctx->odb,
1863                                         ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1864                 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1865
1866                 result = rename(ctx->graph_name, final_graph_name);
1867
1868                 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1869                         fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1870
1871                 if (result) {
1872                         error(_("failed to rename temporary commit-graph file"));
1873                         return -1;
1874                 }
1875         }
1876
1877         commit_lock_file(&lk);
1878
1879         return 0;
1880 }
1881
1882 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1883 {
1884         struct commit_graph *g;
1885         uint32_t num_commits;
1886         enum commit_graph_split_flags flags = COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1887         uint32_t i;
1888
1889         int max_commits = 0;
1890         int size_mult = 2;
1891
1892         if (ctx->split_opts) {
1893                 max_commits = ctx->split_opts->max_commits;
1894
1895                 if (ctx->split_opts->size_multiple)
1896                         size_mult = ctx->split_opts->size_multiple;
1897
1898                 flags = ctx->split_opts->flags;
1899         }
1900
1901         g = ctx->r->objects->commit_graph;
1902         num_commits = ctx->commits.nr;
1903         if (flags == COMMIT_GRAPH_SPLIT_REPLACE)
1904                 ctx->num_commit_graphs_after = 1;
1905         else
1906                 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1907
1908         if (flags != COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED &&
1909             flags != COMMIT_GRAPH_SPLIT_REPLACE) {
1910                 while (g && (g->num_commits <= size_mult * num_commits ||
1911                             (max_commits && num_commits > max_commits))) {
1912                         if (g->odb != ctx->odb)
1913                                 break;
1914
1915                         num_commits += g->num_commits;
1916                         g = g->base_graph;
1917
1918                         ctx->num_commit_graphs_after--;
1919                 }
1920         }
1921
1922         if (flags != COMMIT_GRAPH_SPLIT_REPLACE)
1923                 ctx->new_base_graph = g;
1924         else if (ctx->num_commit_graphs_after != 1)
1925                 BUG("split_graph_merge_strategy: num_commit_graphs_after "
1926                     "should be 1 with --split=replace");
1927
1928         if (ctx->num_commit_graphs_after == 2) {
1929                 char *old_graph_name = get_commit_graph_filename(g->odb);
1930
1931                 if (!strcmp(g->filename, old_graph_name) &&
1932                     g->odb != ctx->odb) {
1933                         ctx->num_commit_graphs_after = 1;
1934                         ctx->new_base_graph = NULL;
1935                 }
1936
1937                 free(old_graph_name);
1938         }
1939
1940         CALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1941         CALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1942
1943         for (i = 0; i < ctx->num_commit_graphs_after &&
1944                     i < ctx->num_commit_graphs_before; i++)
1945                 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1946
1947         i = ctx->num_commit_graphs_before - 1;
1948         g = ctx->r->objects->commit_graph;
1949
1950         while (g) {
1951                 if (i < ctx->num_commit_graphs_after)
1952                         ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1953
1954                 i--;
1955                 g = g->base_graph;
1956         }
1957 }
1958
1959 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1960                                struct commit_graph *g)
1961 {
1962         uint32_t i;
1963         uint32_t offset = g->num_commits_in_base;
1964
1965         ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1966
1967         for (i = 0; i < g->num_commits; i++) {
1968                 struct object_id oid;
1969                 struct commit *result;
1970
1971                 display_progress(ctx->progress, i + 1);
1972
1973                 load_oid_from_graph(g, i + offset, &oid);
1974
1975                 /* only add commits if they still exist in the repo */
1976                 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1977
1978                 if (result) {
1979                         ctx->commits.list[ctx->commits.nr] = result;
1980                         ctx->commits.nr++;
1981                 }
1982         }
1983 }
1984
1985 static int commit_compare(const void *_a, const void *_b)
1986 {
1987         const struct commit *a = *(const struct commit **)_a;
1988         const struct commit *b = *(const struct commit **)_b;
1989         return oidcmp(&a->object.oid, &b->object.oid);
1990 }
1991
1992 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1993 {
1994         uint32_t i;
1995
1996         if (ctx->report_progress)
1997                 ctx->progress = start_delayed_progress(
1998                                         _("Scanning merged commits"),
1999                                         ctx->commits.nr);
2000
2001         QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
2002
2003         ctx->num_extra_edges = 0;
2004         for (i = 0; i < ctx->commits.nr; i++) {
2005                 display_progress(ctx->progress, i);
2006
2007                 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
2008                           &ctx->commits.list[i]->object.oid)) {
2009                         die(_("unexpected duplicate commit id %s"),
2010                             oid_to_hex(&ctx->commits.list[i]->object.oid));
2011                 } else {
2012                         unsigned int num_parents;
2013
2014                         num_parents = commit_list_count(ctx->commits.list[i]->parents);
2015                         if (num_parents > 2)
2016                                 ctx->num_extra_edges += num_parents - 1;
2017                 }
2018         }
2019
2020         stop_progress(&ctx->progress);
2021 }
2022
2023 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
2024 {
2025         struct commit_graph *g = ctx->r->objects->commit_graph;
2026         uint32_t current_graph_number = ctx->num_commit_graphs_before;
2027
2028         while (g && current_graph_number >= ctx->num_commit_graphs_after) {
2029                 current_graph_number--;
2030
2031                 if (ctx->report_progress)
2032                         ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
2033
2034                 merge_commit_graph(ctx, g);
2035                 stop_progress(&ctx->progress);
2036
2037                 g = g->base_graph;
2038         }
2039
2040         if (g) {
2041                 ctx->new_base_graph = g;
2042                 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
2043         }
2044
2045         if (ctx->new_base_graph)
2046                 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
2047
2048         sort_and_scan_merged_commits(ctx);
2049 }
2050
2051 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
2052 {
2053         uint32_t i;
2054         time_t now = time(NULL);
2055
2056         for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
2057                 struct stat st;
2058                 struct utimbuf updated_time;
2059
2060                 stat(ctx->commit_graph_filenames_before[i], &st);
2061
2062                 updated_time.actime = st.st_atime;
2063                 updated_time.modtime = now;
2064                 utime(ctx->commit_graph_filenames_before[i], &updated_time);
2065         }
2066 }
2067
2068 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
2069 {
2070         struct strbuf path = STRBUF_INIT;
2071         DIR *dir;
2072         struct dirent *de;
2073         size_t dirnamelen;
2074         timestamp_t expire_time = time(NULL);
2075
2076         if (ctx->split_opts && ctx->split_opts->expire_time)
2077                 expire_time = ctx->split_opts->expire_time;
2078         if (!ctx->split) {
2079                 char *chain_file_name = get_chain_filename(ctx->odb);
2080                 unlink(chain_file_name);
2081                 free(chain_file_name);
2082                 ctx->num_commit_graphs_after = 0;
2083         }
2084
2085         strbuf_addstr(&path, ctx->odb->path);
2086         strbuf_addstr(&path, "/info/commit-graphs");
2087         dir = opendir(path.buf);
2088
2089         if (!dir)
2090                 goto out;
2091
2092         strbuf_addch(&path, '/');
2093         dirnamelen = path.len;
2094         while ((de = readdir(dir)) != NULL) {
2095                 struct stat st;
2096                 uint32_t i, found = 0;
2097
2098                 strbuf_setlen(&path, dirnamelen);
2099                 strbuf_addstr(&path, de->d_name);
2100
2101                 stat(path.buf, &st);
2102
2103                 if (st.st_mtime > expire_time)
2104                         continue;
2105                 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
2106                         continue;
2107
2108                 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2109                         if (!strcmp(ctx->commit_graph_filenames_after[i],
2110                                     path.buf)) {
2111                                 found = 1;
2112                                 break;
2113                         }
2114                 }
2115
2116                 if (!found)
2117                         unlink(path.buf);
2118         }
2119
2120 out:
2121         strbuf_release(&path);
2122 }
2123
2124 int write_commit_graph(struct object_directory *odb,
2125                        struct string_list *pack_indexes,
2126                        struct oidset *commits,
2127                        enum commit_graph_write_flags flags,
2128                        const struct split_commit_graph_opts *split_opts)
2129 {
2130         struct write_commit_graph_context *ctx;
2131         uint32_t i, count_distinct = 0;
2132         int res = 0;
2133         int replace = 0;
2134         struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
2135
2136         if (!commit_graph_compatible(the_repository))
2137                 return 0;
2138
2139         ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
2140         ctx->r = the_repository;
2141         ctx->odb = odb;
2142         ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
2143         ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
2144         ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
2145         ctx->split_opts = split_opts;
2146         ctx->total_bloom_filter_data_size = 0;
2147
2148         bloom_settings.bits_per_entry = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
2149                                                       bloom_settings.bits_per_entry);
2150         bloom_settings.num_hashes = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
2151                                                   bloom_settings.num_hashes);
2152         bloom_settings.max_changed_paths = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS",
2153                                                          bloom_settings.max_changed_paths);
2154         ctx->bloom_settings = &bloom_settings;
2155
2156         if (flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS)
2157                 ctx->changed_paths = 1;
2158         if (!(flags & COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS)) {
2159                 struct commit_graph *g;
2160                 prepare_commit_graph_one(ctx->r, ctx->odb);
2161
2162                 g = ctx->r->objects->commit_graph;
2163
2164                 /* We have changed-paths already. Keep them in the next graph */
2165                 if (g && g->chunk_bloom_data) {
2166                         ctx->changed_paths = 1;
2167                         ctx->bloom_settings = g->bloom_filter_settings;
2168                 }
2169         }
2170
2171         if (ctx->split) {
2172                 struct commit_graph *g;
2173                 prepare_commit_graph(ctx->r);
2174
2175                 g = ctx->r->objects->commit_graph;
2176
2177                 while (g) {
2178                         ctx->num_commit_graphs_before++;
2179                         g = g->base_graph;
2180                 }
2181
2182                 if (ctx->num_commit_graphs_before) {
2183                         ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2184                         i = ctx->num_commit_graphs_before;
2185                         g = ctx->r->objects->commit_graph;
2186
2187                         while (g) {
2188                                 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2189                                 g = g->base_graph;
2190                         }
2191                 }
2192
2193                 if (ctx->split_opts)
2194                         replace = ctx->split_opts->flags & COMMIT_GRAPH_SPLIT_REPLACE;
2195         }
2196
2197         ctx->approx_nr_objects = approximate_object_count();
2198         ctx->oids.alloc = ctx->approx_nr_objects / 32;
2199
2200         if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2201                 ctx->oids.alloc = split_opts->max_commits;
2202
2203         if (ctx->append) {
2204                 prepare_commit_graph_one(ctx->r, ctx->odb);
2205                 if (ctx->r->objects->commit_graph)
2206                         ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2207         }
2208
2209         if (ctx->oids.alloc < 1024)
2210                 ctx->oids.alloc = 1024;
2211         ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2212
2213         if (ctx->append && ctx->r->objects->commit_graph) {
2214                 struct commit_graph *g = ctx->r->objects->commit_graph;
2215                 for (i = 0; i < g->num_commits; i++) {
2216                         const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2217                         hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2218                 }
2219         }
2220
2221         if (pack_indexes) {
2222                 ctx->order_by_pack = 1;
2223                 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2224                         goto cleanup;
2225         }
2226
2227         if (commits) {
2228                 if ((res = fill_oids_from_commits(ctx, commits)))
2229                         goto cleanup;
2230         }
2231
2232         if (!pack_indexes && !commits) {
2233                 ctx->order_by_pack = 1;
2234                 fill_oids_from_all_packs(ctx);
2235         }
2236
2237         close_reachable(ctx);
2238
2239         count_distinct = count_distinct_commits(ctx);
2240
2241         if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2242                 error(_("the commit graph format cannot write %d commits"), count_distinct);
2243                 res = -1;
2244                 goto cleanup;
2245         }
2246
2247         ctx->commits.alloc = count_distinct;
2248         ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2249
2250         copy_oids_to_commits(ctx);
2251
2252         if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2253                 error(_("too many commits to write graph"));
2254                 res = -1;
2255                 goto cleanup;
2256         }
2257
2258         if (!ctx->commits.nr && !replace)
2259                 goto cleanup;
2260
2261         if (ctx->split) {
2262                 split_graph_merge_strategy(ctx);
2263
2264                 if (!replace)
2265                         merge_commit_graphs(ctx);
2266         } else
2267                 ctx->num_commit_graphs_after = 1;
2268
2269         compute_generation_numbers(ctx);
2270
2271         if (ctx->changed_paths)
2272                 compute_bloom_filters(ctx);
2273
2274         res = write_commit_graph_file(ctx);
2275
2276         if (ctx->split)
2277                 mark_commit_graphs(ctx);
2278
2279         expire_commit_graphs(ctx);
2280
2281 cleanup:
2282         free(ctx->graph_name);
2283         free(ctx->commits.list);
2284         free(ctx->oids.list);
2285
2286         if (ctx->commit_graph_filenames_after) {
2287                 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2288                         free(ctx->commit_graph_filenames_after[i]);
2289                         free(ctx->commit_graph_hash_after[i]);
2290                 }
2291
2292                 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2293                         free(ctx->commit_graph_filenames_before[i]);
2294
2295                 free(ctx->commit_graph_filenames_after);
2296                 free(ctx->commit_graph_filenames_before);
2297                 free(ctx->commit_graph_hash_after);
2298         }
2299
2300         free(ctx);
2301
2302         return res;
2303 }
2304
2305 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2306 static int verify_commit_graph_error;
2307
2308 static void graph_report(const char *fmt, ...)
2309 {
2310         va_list ap;
2311
2312         verify_commit_graph_error = 1;
2313         va_start(ap, fmt);
2314         vfprintf(stderr, fmt, ap);
2315         fprintf(stderr, "\n");
2316         va_end(ap);
2317 }
2318
2319 #define GENERATION_ZERO_EXISTS 1
2320 #define GENERATION_NUMBER_EXISTS 2
2321
2322 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2323 {
2324         uint32_t i, cur_fanout_pos = 0;
2325         struct object_id prev_oid, cur_oid, checksum;
2326         int generation_zero = 0;
2327         struct hashfile *f;
2328         int devnull;
2329         struct progress *progress = NULL;
2330         int local_error = 0;
2331
2332         if (!g) {
2333                 graph_report("no commit-graph file loaded");
2334                 return 1;
2335         }
2336
2337         verify_commit_graph_error = verify_commit_graph_lite(g);
2338         if (verify_commit_graph_error)
2339                 return verify_commit_graph_error;
2340
2341         devnull = open("/dev/null", O_WRONLY);
2342         f = hashfd(devnull, NULL);
2343         hashwrite(f, g->data, g->data_len - g->hash_len);
2344         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2345         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2346                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2347                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2348         }
2349
2350         for (i = 0; i < g->num_commits; i++) {
2351                 struct commit *graph_commit;
2352
2353                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2354
2355                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2356                         graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2357                                      oid_to_hex(&prev_oid),
2358                                      oid_to_hex(&cur_oid));
2359
2360                 oidcpy(&prev_oid, &cur_oid);
2361
2362                 while (cur_oid.hash[0] > cur_fanout_pos) {
2363                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2364
2365                         if (i != fanout_value)
2366                                 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2367                                              cur_fanout_pos, fanout_value, i);
2368                         cur_fanout_pos++;
2369                 }
2370
2371                 graph_commit = lookup_commit(r, &cur_oid);
2372                 if (!parse_commit_in_graph_one(r, g, graph_commit))
2373                         graph_report(_("failed to parse commit %s from commit-graph"),
2374                                      oid_to_hex(&cur_oid));
2375         }
2376
2377         while (cur_fanout_pos < 256) {
2378                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2379
2380                 if (g->num_commits != fanout_value)
2381                         graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2382                                      cur_fanout_pos, fanout_value, i);
2383
2384                 cur_fanout_pos++;
2385         }
2386
2387         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2388                 return verify_commit_graph_error;
2389
2390         if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2391                 progress = start_progress(_("Verifying commits in commit graph"),
2392                                         g->num_commits);
2393
2394         for (i = 0; i < g->num_commits; i++) {
2395                 struct commit *graph_commit, *odb_commit;
2396                 struct commit_list *graph_parents, *odb_parents;
2397                 uint32_t max_generation = 0;
2398                 uint32_t generation;
2399
2400                 display_progress(progress, i + 1);
2401                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2402
2403                 graph_commit = lookup_commit(r, &cur_oid);
2404                 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2405                 if (parse_commit_internal(odb_commit, 0, 0)) {
2406                         graph_report(_("failed to parse commit %s from object database for commit-graph"),
2407                                      oid_to_hex(&cur_oid));
2408                         continue;
2409                 }
2410
2411                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2412                            get_commit_tree_oid(odb_commit)))
2413                         graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2414                                      oid_to_hex(&cur_oid),
2415                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
2416                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
2417
2418                 graph_parents = graph_commit->parents;
2419                 odb_parents = odb_commit->parents;
2420
2421                 while (graph_parents) {
2422                         if (odb_parents == NULL) {
2423                                 graph_report(_("commit-graph parent list for commit %s is too long"),
2424                                              oid_to_hex(&cur_oid));
2425                                 break;
2426                         }
2427
2428                         /* parse parent in case it is in a base graph */
2429                         parse_commit_in_graph_one(r, g, graph_parents->item);
2430
2431                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2432                                 graph_report(_("commit-graph parent for %s is %s != %s"),
2433                                              oid_to_hex(&cur_oid),
2434                                              oid_to_hex(&graph_parents->item->object.oid),
2435                                              oid_to_hex(&odb_parents->item->object.oid));
2436
2437                         generation = commit_graph_generation(graph_parents->item);
2438                         if (generation > max_generation)
2439                                 max_generation = generation;
2440
2441                         graph_parents = graph_parents->next;
2442                         odb_parents = odb_parents->next;
2443                 }
2444
2445                 if (odb_parents != NULL)
2446                         graph_report(_("commit-graph parent list for commit %s terminates early"),
2447                                      oid_to_hex(&cur_oid));
2448
2449                 if (!commit_graph_generation(graph_commit)) {
2450                         if (generation_zero == GENERATION_NUMBER_EXISTS)
2451                                 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2452                                              oid_to_hex(&cur_oid));
2453                         generation_zero = GENERATION_ZERO_EXISTS;
2454                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2455                         graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2456                                      oid_to_hex(&cur_oid));
2457
2458                 if (generation_zero == GENERATION_ZERO_EXISTS)
2459                         continue;
2460
2461                 /*
2462                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
2463                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2464                  * extra logic in the following condition.
2465                  */
2466                 if (max_generation == GENERATION_NUMBER_MAX)
2467                         max_generation--;
2468
2469                 generation = commit_graph_generation(graph_commit);
2470                 if (generation != max_generation + 1)
2471                         graph_report(_("commit-graph generation for commit %s is %u != %u"),
2472                                      oid_to_hex(&cur_oid),
2473                                      generation,
2474                                      max_generation + 1);
2475
2476                 if (graph_commit->date != odb_commit->date)
2477                         graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2478                                      oid_to_hex(&cur_oid),
2479                                      graph_commit->date,
2480                                      odb_commit->date);
2481         }
2482         stop_progress(&progress);
2483
2484         local_error = verify_commit_graph_error;
2485
2486         if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2487                 local_error |= verify_commit_graph(r, g->base_graph, flags);
2488
2489         return local_error;
2490 }
2491
2492 void free_commit_graph(struct commit_graph *g)
2493 {
2494         if (!g)
2495                 return;
2496         if (g->data) {
2497                 munmap((void *)g->data, g->data_len);
2498                 g->data = NULL;
2499         }
2500         free(g->filename);
2501         free(g->bloom_filter_settings);
2502         free(g);
2503 }
2504
2505 void disable_commit_graph(struct repository *r)
2506 {
2507         r->commit_graph_disabled = 1;
2508 }