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
 970 static int write_graph_chunk_fanout(struct hashfile *f,
 971                                     struct write_commit_graph_context *ctx)
 972 {
 973         int i, count = 0;
 974         struct commit **list = ctx->commits.list;
 975
 976         /*
 977          * Write the first-level table (the list is sorted,
 978          * but we use a 256-entry lookup to be able to avoid
 979          * having to do eight extra binary search iterations).
 980          */
 981         for (i = 0; i < 256; i++) {
 982                 while (count < ctx->commits.nr) {
 983                         if ((*list)->object.oid.hash[0] != i)
 984                                 break;
 985                         display_progress(ctx->progress, ++ctx->progress_cnt);
 986                         count++;
 987                         list++;
 988                 }
 989
 990                 hashwrite_be32(f, count);
 991         }
 992
 993         return 0;
 994 }
 995
 996 static int write_graph_chunk_oids(struct hashfile *f,
 997                                   struct write_commit_graph_context *ctx)
 998 {
 999         struct commit **list = ctx->commits.list;
1000         int count;
1001         for (count = 0; count < ctx->commits.nr; count++, list++) {
1002                 display_progress(ctx->progress, ++ctx->progress_cnt);
1003                 hashwrite(f, (*list)->object.oid.hash, the_hash_algo->rawsz);
1004         }
1005
1006         return 0;
1007 }
1008
1009 static const unsigned char *commit_to_sha1(size_t index, void *table)
1010 {
1011         struct commit **commits = table;
1012         return commits[index]->object.oid.hash;
1013 }
1014
1015 static int write_graph_chunk_data(struct hashfile *f,
1016                                   struct write_commit_graph_context *ctx)
1017 {
1018         struct commit **list = ctx->commits.list;
1019         struct commit **last = ctx->commits.list + ctx->commits.nr;
1020         uint32_t num_extra_edges = 0;
1021
1022         while (list < last) {
1023                 struct commit_list *parent;
1024                 struct object_id *tree;
1025                 int edge_value;
1026                 uint32_t packedDate[2];
1027                 display_progress(ctx->progress, ++ctx->progress_cnt);
1028
1029                 if (parse_commit_no_graph(*list))
1030                         die(_("unable to parse commit %s"),
1031                                 oid_to_hex(&(*list)->object.oid));
1032                 tree = get_commit_tree_oid(*list);
1033                 hashwrite(f, tree->hash, the_hash_algo->rawsz);
1034
1035                 parent = (*list)->parents;
1036
1037                 if (!parent)
1038                         edge_value = GRAPH_PARENT_NONE;
1039                 else {
1040                         edge_value = sha1_pos(parent->item->object.oid.hash,
1041                                               ctx->commits.list,
1042                                               ctx->commits.nr,
1043                                               commit_to_sha1);
1044
1045                         if (edge_value >= 0)
1046                                 edge_value += ctx->new_num_commits_in_base;
1047                         else if (ctx->new_base_graph) {
1048                                 uint32_t pos;
1049                                 if (find_commit_in_graph(parent->item,
1050                                                          ctx->new_base_graph,
1051                                                          &pos))
1052                                         edge_value = pos;
1053                         }
1054
1055                         if (edge_value < 0)
1056                                 BUG("missing parent %s for commit %s",
1057                                     oid_to_hex(&parent->item->object.oid),
1058                                     oid_to_hex(&(*list)->object.oid));
1059                 }
1060
1061                 hashwrite_be32(f, edge_value);
1062
1063                 if (parent)
1064                         parent = parent->next;
1065
1066                 if (!parent)
1067                         edge_value = GRAPH_PARENT_NONE;
1068                 else if (parent->next)
1069                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
1070                 else {
1071                         edge_value = sha1_pos(parent->item->object.oid.hash,
1072                                               ctx->commits.list,
1073                                               ctx->commits.nr,
1074                                               commit_to_sha1);
1075
1076                         if (edge_value >= 0)
1077                                 edge_value += ctx->new_num_commits_in_base;
1078                         else if (ctx->new_base_graph) {
1079                                 uint32_t pos;
1080                                 if (find_commit_in_graph(parent->item,
1081                                                          ctx->new_base_graph,
1082                                                          &pos))
1083                                         edge_value = pos;
1084                         }
1085
1086                         if (edge_value < 0)
1087                                 BUG("missing parent %s for commit %s",
1088                                     oid_to_hex(&parent->item->object.oid),
1089                                     oid_to_hex(&(*list)->object.oid));
1090                 }
1091
1092                 hashwrite_be32(f, edge_value);
1093
1094                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
1095                         do {
1096                                 num_extra_edges++;
1097                                 parent = parent->next;
1098                         } while (parent);
1099                 }
1100
1101                 if (sizeof((*list)->date) > 4)
1102                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
1103                 else
1104                         packedDate[0] = 0;
1105
1106                 packedDate[0] |= htonl(commit_graph_data_at(*list)->generation << 2);
1107
1108                 packedDate[1] = htonl((*list)->date);
1109                 hashwrite(f, packedDate, 8);
1110
1111                 list++;
1112         }
1113
1114         return 0;
1115 }
1116
1117 static int write_graph_chunk_extra_edges(struct hashfile *f,
1118                                          struct write_commit_graph_context *ctx)
1119 {
1120         struct commit **list = ctx->commits.list;
1121         struct commit **last = ctx->commits.list + ctx->commits.nr;
1122         struct commit_list *parent;
1123
1124         while (list < last) {
1125                 int num_parents = 0;
1126
1127                 display_progress(ctx->progress, ++ctx->progress_cnt);
1128
1129                 for (parent = (*list)->parents; num_parents < 3 && parent;
1130                      parent = parent->next)
1131                         num_parents++;
1132
1133                 if (num_parents <= 2) {
1134                         list++;
1135                         continue;
1136                 }
1137
1138                 /* Since num_parents > 2, this initializer is safe. */
1139                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
1140                         int edge_value = sha1_pos(parent->item->object.oid.hash,
1141                                                   ctx->commits.list,
1142                                                   ctx->commits.nr,
1143                                                   commit_to_sha1);
1144
1145                         if (edge_value >= 0)
1146                                 edge_value += ctx->new_num_commits_in_base;
1147                         else if (ctx->new_base_graph) {
1148                                 uint32_t pos;
1149                                 if (find_commit_in_graph(parent->item,
1150                                                          ctx->new_base_graph,
1151                                                          &pos))
1152                                         edge_value = pos;
1153                         }
1154
1155                         if (edge_value < 0)
1156                                 BUG("missing parent %s for commit %s",
1157                                     oid_to_hex(&parent->item->object.oid),
1158                                     oid_to_hex(&(*list)->object.oid));
1159                         else if (!parent->next)
1160                                 edge_value |= GRAPH_LAST_EDGE;
1161
1162                         hashwrite_be32(f, edge_value);
1163                 }
1164
1165                 list++;
1166         }
1167
1168         return 0;
1169 }
1170
1171 static int write_graph_chunk_bloom_indexes(struct hashfile *f,
1172                                            struct write_commit_graph_context *ctx)
1173 {
1174         struct commit **list = ctx->commits.list;
1175         struct commit **last = ctx->commits.list + ctx->commits.nr;
1176         uint32_t cur_pos = 0;
1177
1178         while (list < last) {
1179                 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1180                 size_t len = filter ? filter->len : 0;
1181                 cur_pos += len;
1182                 display_progress(ctx->progress, ++ctx->progress_cnt);
1183                 hashwrite_be32(f, cur_pos);
1184                 list++;
1185         }
1186
1187         return 0;
1188 }
1189
1190 static void trace2_bloom_filter_settings(struct write_commit_graph_context *ctx)
1191 {
1192         struct json_writer jw = JSON_WRITER_INIT;
1193
1194         jw_object_begin(&jw, 0);
1195         jw_object_intmax(&jw, "hash_version", ctx->bloom_settings->hash_version);
1196         jw_object_intmax(&jw, "num_hashes", ctx->bloom_settings->num_hashes);
1197         jw_object_intmax(&jw, "bits_per_entry", ctx->bloom_settings->bits_per_entry);
1198         jw_object_intmax(&jw, "max_changed_paths", ctx->bloom_settings->max_changed_paths);
1199         jw_end(&jw);
1200
1201         trace2_data_json("bloom", ctx->r, "settings", &jw);
1202
1203         jw_release(&jw);
1204 }
1205
1206 static int write_graph_chunk_bloom_data(struct hashfile *f,
1207                                         struct write_commit_graph_context *ctx)
1208 {
1209         struct commit **list = ctx->commits.list;
1210         struct commit **last = ctx->commits.list + ctx->commits.nr;
1211
1212         trace2_bloom_filter_settings(ctx);
1213
1214         hashwrite_be32(f, ctx->bloom_settings->hash_version);
1215         hashwrite_be32(f, ctx->bloom_settings->num_hashes);
1216         hashwrite_be32(f, ctx->bloom_settings->bits_per_entry);
1217
1218         while (list < last) {
1219                 struct bloom_filter *filter = get_bloom_filter(ctx->r, *list, 0);
1220                 size_t len = filter ? filter->len : 0;
1221
1222                 display_progress(ctx->progress, ++ctx->progress_cnt);
1223                 if (len)
1224                         hashwrite(f, filter->data, len * sizeof(unsigned char));
1225                 list++;
1226         }
1227
1228         return 0;
1229 }
1230
1231 static int oid_compare(const void *_a, const void *_b)
1232 {
1233         const struct object_id *a = (const struct object_id *)_a;
1234         const struct object_id *b = (const struct object_id *)_b;
1235         return oidcmp(a, b);
1236 }
1237
1238 static int add_packed_commits(const struct object_id *oid,
1239                               struct packed_git *pack,
1240                               uint32_t pos,
1241                               void *data)
1242 {
1243         struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
1244         enum object_type type;
1245         off_t offset = nth_packed_object_offset(pack, pos);
1246         struct object_info oi = OBJECT_INFO_INIT;
1247
1248         if (ctx->progress)
1249                 display_progress(ctx->progress, ++ctx->progress_done);
1250
1251         oi.typep = &type;
1252         if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
1253                 die(_("unable to get type of object %s"), oid_to_hex(oid));
1254
1255         if (type != OBJ_COMMIT)
1256                 return 0;
1257
1258         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1259         oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
1260         ctx->oids.nr++;
1261
1262         set_commit_pos(ctx->r, oid);
1263
1264         return 0;
1265 }
1266
1267 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
1268 {
1269         struct commit_list *parent;
1270         for (parent = commit->parents; parent; parent = parent->next) {
1271                 if (!(parent->item->object.flags & REACHABLE)) {
1272                         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1273                         oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
1274                         ctx->oids.nr++;
1275                         parent->item->object.flags |= REACHABLE;
1276                 }
1277         }
1278 }
1279
1280 static void close_reachable(struct write_commit_graph_context *ctx)
1281 {
1282         int i;
1283         struct commit *commit;
1284         enum commit_graph_split_flags flags = ctx->split_opts ?
1285                 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1286
1287         if (ctx->report_progress)
1288                 ctx->progress = start_delayed_progress(
1289                                         _("Loading known commits in commit graph"),
1290                                         ctx->oids.nr);
1291         for (i = 0; i < ctx->oids.nr; i++) {
1292                 display_progress(ctx->progress, i + 1);
1293                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1294                 if (commit)
1295                         commit->object.flags |= REACHABLE;
1296         }
1297         stop_progress(&ctx->progress);
1298
1299         /*
1300          * As this loop runs, ctx->oids.nr may grow, but not more
1301          * than the number of missing commits in the reachable
1302          * closure.
1303          */
1304         if (ctx->report_progress)
1305                 ctx->progress = start_delayed_progress(
1306                                         _("Expanding reachable commits in commit graph"),
1307                                         0);
1308         for (i = 0; i < ctx->oids.nr; i++) {
1309                 display_progress(ctx->progress, i + 1);
1310                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1311
1312                 if (!commit)
1313                         continue;
1314                 if (ctx->split) {
1315                         if ((!parse_commit(commit) &&
1316                              commit_graph_position(commit) == COMMIT_NOT_FROM_GRAPH) ||
1317                             flags == COMMIT_GRAPH_SPLIT_REPLACE)
1318                                 add_missing_parents(ctx, commit);
1319                 } else if (!parse_commit_no_graph(commit))
1320                         add_missing_parents(ctx, commit);
1321         }
1322         stop_progress(&ctx->progress);
1323
1324         if (ctx->report_progress)
1325                 ctx->progress = start_delayed_progress(
1326                                         _("Clearing commit marks in commit graph"),
1327                                         ctx->oids.nr);
1328         for (i = 0; i < ctx->oids.nr; i++) {
1329                 display_progress(ctx->progress, i + 1);
1330                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
1331
1332                 if (commit)
1333                         commit->object.flags &= ~REACHABLE;
1334         }
1335         stop_progress(&ctx->progress);
1336 }
1337
1338 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
1339 {
1340         int i;
1341         struct commit_list *list = NULL;
1342
1343         if (ctx->report_progress)
1344                 ctx->progress = start_delayed_progress(
1345                                         _("Computing commit graph generation numbers"),
1346                                         ctx->commits.nr);
1347         for (i = 0; i < ctx->commits.nr; i++) {
1348                 uint32_t generation = commit_graph_data_at(ctx->commits.list[i])->generation;
1349
1350                 display_progress(ctx->progress, i + 1);
1351                 if (generation != GENERATION_NUMBER_INFINITY &&
1352                     generation != GENERATION_NUMBER_ZERO)
1353                         continue;
1354
1355                 commit_list_insert(ctx->commits.list[i], &list);
1356                 while (list) {
1357                         struct commit *current = list->item;
1358                         struct commit_list *parent;
1359                         int all_parents_computed = 1;
1360                         uint32_t max_generation = 0;
1361
1362                         for (parent = current->parents; parent; parent = parent->next) {
1363                                 generation = commit_graph_data_at(parent->item)->generation;
1364
1365                                 if (generation == GENERATION_NUMBER_INFINITY ||
1366                                     generation == GENERATION_NUMBER_ZERO) {
1367                                         all_parents_computed = 0;
1368                                         commit_list_insert(parent->item, &list);
1369                                         break;
1370                                 } else if (generation > max_generation) {
1371                                         max_generation = generation;
1372                                 }
1373                         }
1374
1375                         if (all_parents_computed) {
1376                                 struct commit_graph_data *data = commit_graph_data_at(current);
1377
1378                                 data->generation = max_generation + 1;
1379                                 pop_commit(&list);
1380
1381                                 if (data->generation > GENERATION_NUMBER_MAX)
1382                                         data->generation = GENERATION_NUMBER_MAX;
1383                         }
1384                 }
1385         }
1386         stop_progress(&ctx->progress);
1387 }
1388
1389 static void compute_bloom_filters(struct write_commit_graph_context *ctx)
1390 {
1391         int i;
1392         struct progress *progress = NULL;
1393         struct commit **sorted_commits;
1394
1395         init_bloom_filters();
1396
1397         if (ctx->report_progress)
1398                 progress = start_delayed_progress(
1399                         _("Computing commit changed paths Bloom filters"),
1400                         ctx->commits.nr);
1401
1402         ALLOC_ARRAY(sorted_commits, ctx->commits.nr);
1403         COPY_ARRAY(sorted_commits, ctx->commits.list, ctx->commits.nr);
1404
1405         if (ctx->order_by_pack)
1406                 QSORT(sorted_commits, ctx->commits.nr, commit_pos_cmp);
1407         else
1408                 QSORT(sorted_commits, ctx->commits.nr, commit_gen_cmp);
1409
1410         for (i = 0; i < ctx->commits.nr; i++) {
1411                 struct commit *c = sorted_commits[i];
1412                 struct bloom_filter *filter = get_bloom_filter(ctx->r, c, 1);
1413                 ctx->total_bloom_filter_data_size += sizeof(unsigned char) * filter->len;
1414                 display_progress(progress, i + 1);
1415         }
1416
1417         free(sorted_commits);
1418         stop_progress(&progress);
1419 }
1420
1421 struct refs_cb_data {
1422         struct oidset *commits;
1423         struct progress *progress;
1424 };
1425
1426 static int add_ref_to_set(const char *refname,
1427                           const struct object_id *oid,
1428                           int flags, void *cb_data)
1429 {
1430         struct object_id peeled;
1431         struct refs_cb_data *data = (struct refs_cb_data *)cb_data;
1432
1433         if (!peel_ref(refname, &peeled))
1434                 oid = &peeled;
1435         if (oid_object_info(the_repository, oid, NULL) == OBJ_COMMIT)
1436                 oidset_insert(data->commits, oid);
1437
1438         display_progress(data->progress, oidset_size(data->commits));
1439
1440         return 0;
1441 }
1442
1443 int write_commit_graph_reachable(struct object_directory *odb,
1444                                  enum commit_graph_write_flags flags,
1445                                  const struct split_commit_graph_opts *split_opts)
1446 {
1447         struct oidset commits = OIDSET_INIT;
1448         struct refs_cb_data data;
1449         int result;
1450
1451         memset(&data, 0, sizeof(data));
1452         data.commits = &commits;
1453         if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
1454                 data.progress = start_delayed_progress(
1455                         _("Collecting referenced commits"), 0);
1456
1457         for_each_ref(add_ref_to_set, &data);
1458
1459         stop_progress(&data.progress);
1460
1461         result = write_commit_graph(odb, NULL, &commits,
1462                                     flags, split_opts);
1463
1464         oidset_clear(&commits);
1465         return result;
1466 }
1467
1468 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1469                                 struct string_list *pack_indexes)
1470 {
1471         uint32_t i;
1472         struct strbuf progress_title = STRBUF_INIT;
1473         struct strbuf packname = STRBUF_INIT;
1474         int dirlen;
1475
1476         strbuf_addf(&packname, "%s/pack/", ctx->odb->path);
1477         dirlen = packname.len;
1478         if (ctx->report_progress) {
1479                 strbuf_addf(&progress_title,
1480                             Q_("Finding commits for commit graph in %d pack",
1481                                "Finding commits for commit graph in %d packs",
1482                                pack_indexes->nr),
1483                             pack_indexes->nr);
1484                 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1485                 ctx->progress_done = 0;
1486         }
1487         for (i = 0; i < pack_indexes->nr; i++) {
1488                 struct packed_git *p;
1489                 strbuf_setlen(&packname, dirlen);
1490                 strbuf_addstr(&packname, pack_indexes->items[i].string);
1491                 p = add_packed_git(packname.buf, packname.len, 1);
1492                 if (!p) {
1493                         error(_("error adding pack %s"), packname.buf);
1494                         return -1;
1495                 }
1496                 if (open_pack_index(p)) {
1497                         error(_("error opening index for %s"), packname.buf);
1498                         return -1;
1499                 }
1500                 for_each_object_in_pack(p, add_packed_commits, ctx,
1501                                         FOR_EACH_OBJECT_PACK_ORDER);
1502                 close_pack(p);
1503                 free(p);
1504         }
1505
1506         stop_progress(&ctx->progress);
1507         strbuf_release(&progress_title);
1508         strbuf_release(&packname);
1509
1510         return 0;
1511 }
1512
1513 static int fill_oids_from_commits(struct write_commit_graph_context *ctx,
1514                                   struct oidset *commits)
1515 {
1516         struct oidset_iter iter;
1517         struct object_id *oid;
1518
1519         if (!oidset_size(commits))
1520                 return 0;
1521
1522         oidset_iter_init(commits, &iter);
1523         while ((oid = oidset_iter_next(&iter))) {
1524                 ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1525                 oidcpy(&ctx->oids.list[ctx->oids.nr], oid);
1526                 ctx->oids.nr++;
1527         }
1528
1529         return 0;
1530 }
1531
1532 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1533 {
1534         if (ctx->report_progress)
1535                 ctx->progress = start_delayed_progress(
1536                         _("Finding commits for commit graph among packed objects"),
1537                         ctx->approx_nr_objects);
1538         for_each_packed_object(add_packed_commits, ctx,
1539                                FOR_EACH_OBJECT_PACK_ORDER);
1540         if (ctx->progress_done < ctx->approx_nr_objects)
1541                 display_progress(ctx->progress, ctx->approx_nr_objects);
1542         stop_progress(&ctx->progress);
1543 }
1544
1545 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1546 {
1547         uint32_t i, count_distinct = 1;
1548
1549         if (ctx->report_progress)
1550                 ctx->progress = start_delayed_progress(
1551                         _("Counting distinct commits in commit graph"),
1552                         ctx->oids.nr);
1553         display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1554         QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1555
1556         for (i = 1; i < ctx->oids.nr; i++) {
1557                 display_progress(ctx->progress, i + 1);
1558                 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i])) {
1559                         if (ctx->split) {
1560                                 struct commit *c = lookup_commit(ctx->r, &ctx->oids.list[i]);
1561
1562                                 if (!c || commit_graph_position(c) != COMMIT_NOT_FROM_GRAPH)
1563                                         continue;
1564                         }
1565
1566                         count_distinct++;
1567                 }
1568         }
1569         stop_progress(&ctx->progress);
1570
1571         return count_distinct;
1572 }
1573
1574 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1575 {
1576         uint32_t i;
1577         enum commit_graph_split_flags flags = ctx->split_opts ?
1578                 ctx->split_opts->flags : COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1579
1580         ctx->num_extra_edges = 0;
1581         if (ctx->report_progress)
1582                 ctx->progress = start_delayed_progress(
1583                         _("Finding extra edges in commit graph"),
1584                         ctx->oids.nr);
1585         for (i = 0; i < ctx->oids.nr; i++) {
1586                 unsigned int num_parents;
1587
1588                 display_progress(ctx->progress, i + 1);
1589                 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1590                         continue;
1591
1592                 ALLOC_GROW(ctx->commits.list, ctx->commits.nr + 1, ctx->commits.alloc);
1593                 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1594
1595                 if (ctx->split && flags != COMMIT_GRAPH_SPLIT_REPLACE &&
1596                     commit_graph_position(ctx->commits.list[ctx->commits.nr]) != COMMIT_NOT_FROM_GRAPH)
1597                         continue;
1598
1599                 if (ctx->split && flags == COMMIT_GRAPH_SPLIT_REPLACE)
1600                         parse_commit(ctx->commits.list[ctx->commits.nr]);
1601                 else
1602                         parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1603
1604                 num_parents = commit_list_count(ctx->commits.list[ctx->commits.nr]->parents);
1605                 if (num_parents > 2)
1606                         ctx->num_extra_edges += num_parents - 1;
1607
1608                 ctx->commits.nr++;
1609         }
1610         stop_progress(&ctx->progress);
1611 }
1612
1613 static int write_graph_chunk_base_1(struct hashfile *f,
1614                                     struct commit_graph *g)
1615 {
1616         int num = 0;
1617
1618         if (!g)
1619                 return 0;
1620
1621         num = write_graph_chunk_base_1(f, g->base_graph);
1622         hashwrite(f, g->oid.hash, the_hash_algo->rawsz);
1623         return num + 1;
1624 }
1625
1626 static int write_graph_chunk_base(struct hashfile *f,
1627                                   struct write_commit_graph_context *ctx)
1628 {
1629         int num = write_graph_chunk_base_1(f, ctx->new_base_graph);
1630
1631         if (num != ctx->num_commit_graphs_after - 1) {
1632                 error(_("failed to write correct number of base graph ids"));
1633                 return -1;
1634         }
1635
1636         return 0;
1637 }
1638
1639 typedef int (*chunk_write_fn)(struct hashfile *f,
1640                               struct write_commit_graph_context *ctx);
1641
1642 struct chunk_info {
1643         uint32_t id;
1644         uint64_t size;
1645         chunk_write_fn write_fn;
1646 };
1647
1648 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1649 {
1650         uint32_t i;
1651         int fd;
1652         struct hashfile *f;
1653         struct lock_file lk = LOCK_INIT;
1654         struct chunk_info chunks[MAX_NUM_CHUNKS + 1];
1655         const unsigned hashsz = the_hash_algo->rawsz;
1656         struct strbuf progress_title = STRBUF_INIT;
1657         int num_chunks = 3;
1658         uint64_t chunk_offset;
1659         struct object_id file_hash;
1660         struct bloom_filter_settings bloom_settings = DEFAULT_BLOOM_FILTER_SETTINGS;
1661
1662         if (!ctx->bloom_settings) {
1663                 bloom_settings.bits_per_entry = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_BITS_PER_ENTRY",
1664                                                               bloom_settings.bits_per_entry);
1665                 bloom_settings.num_hashes = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_NUM_HASHES",
1666                                                           bloom_settings.num_hashes);
1667                 bloom_settings.max_changed_paths = git_env_ulong("GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS",
1668                                                           bloom_settings.max_changed_paths);
1669                 ctx->bloom_settings = &bloom_settings;
1670         }
1671
1672         if (ctx->split) {
1673                 struct strbuf tmp_file = STRBUF_INIT;
1674
1675                 strbuf_addf(&tmp_file,
1676                             "%s/info/commit-graphs/tmp_graph_XXXXXX",
1677                             ctx->odb->path);
1678                 ctx->graph_name = strbuf_detach(&tmp_file, NULL);
1679         } else {
1680                 ctx->graph_name = get_commit_graph_filename(ctx->odb);
1681         }
1682
1683         if (safe_create_leading_directories(ctx->graph_name)) {
1684                 UNLEAK(ctx->graph_name);
1685                 error(_("unable to create leading directories of %s"),
1686                         ctx->graph_name);
1687                 return -1;
1688         }
1689
1690         if (ctx->split) {
1691                 char *lock_name = get_chain_filename(ctx->odb);
1692
1693                 hold_lock_file_for_update_mode(&lk, lock_name,
1694                                                LOCK_DIE_ON_ERROR, 0444);
1695
1696                 fd = git_mkstemp_mode(ctx->graph_name, 0444);
1697                 if (fd < 0) {
1698                         error(_("unable to create temporary graph layer"));
1699                         return -1;
1700                 }
1701
1702                 if (adjust_shared_perm(ctx->graph_name)) {
1703                         error(_("unable to adjust shared permissions for '%s'"),
1704                               ctx->graph_name);
1705                         return -1;
1706                 }
1707
1708                 f = hashfd(fd, ctx->graph_name);
1709         } else {
1710                 hold_lock_file_for_update_mode(&lk, ctx->graph_name,
1711                                                LOCK_DIE_ON_ERROR, 0444);
1712                 fd = lk.tempfile->fd;
1713                 f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1714         }
1715
1716         chunks[0].id = GRAPH_CHUNKID_OIDFANOUT;
1717         chunks[0].size = GRAPH_FANOUT_SIZE;
1718         chunks[0].write_fn = write_graph_chunk_fanout;
1719         chunks[1].id = GRAPH_CHUNKID_OIDLOOKUP;
1720         chunks[1].size = hashsz * ctx->commits.nr;
1721         chunks[1].write_fn = write_graph_chunk_oids;
1722         chunks[2].id = GRAPH_CHUNKID_DATA;
1723         chunks[2].size = (hashsz + 16) * ctx->commits.nr;
1724         chunks[2].write_fn = write_graph_chunk_data;
1725         if (ctx->num_extra_edges) {
1726                 chunks[num_chunks].id = GRAPH_CHUNKID_EXTRAEDGES;
1727                 chunks[num_chunks].size = 4 * ctx->num_extra_edges;
1728                 chunks[num_chunks].write_fn = write_graph_chunk_extra_edges;
1729                 num_chunks++;
1730         }
1731         if (ctx->changed_paths) {
1732                 chunks[num_chunks].id = GRAPH_CHUNKID_BLOOMINDEXES;
1733                 chunks[num_chunks].size = sizeof(uint32_t) * ctx->commits.nr;
1734                 chunks[num_chunks].write_fn = write_graph_chunk_bloom_indexes;
1735                 num_chunks++;
1736                 chunks[num_chunks].id = GRAPH_CHUNKID_BLOOMDATA;
1737                 chunks[num_chunks].size = sizeof(uint32_t) * 3
1738                                           + ctx->total_bloom_filter_data_size;
1739                 chunks[num_chunks].write_fn = write_graph_chunk_bloom_data;
1740                 num_chunks++;
1741         }
1742         if (ctx->num_commit_graphs_after > 1) {
1743                 chunks[num_chunks].id = GRAPH_CHUNKID_BASE;
1744                 chunks[num_chunks].size = hashsz * (ctx->num_commit_graphs_after - 1);
1745                 chunks[num_chunks].write_fn = write_graph_chunk_base;
1746                 num_chunks++;
1747         }
1748
1749         chunks[num_chunks].id = 0;
1750         chunks[num_chunks].size = 0;
1751
1752         hashwrite_be32(f, GRAPH_SIGNATURE);
1753
1754         hashwrite_u8(f, GRAPH_VERSION);
1755         hashwrite_u8(f, oid_version());
1756         hashwrite_u8(f, num_chunks);
1757         hashwrite_u8(f, ctx->num_commit_graphs_after - 1);
1758
1759         chunk_offset = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1760         for (i = 0; i <= num_chunks; i++) {
1761                 uint32_t chunk_write[3];
1762
1763                 chunk_write[0] = htonl(chunks[i].id);
1764                 chunk_write[1] = htonl(chunk_offset >> 32);
1765                 chunk_write[2] = htonl(chunk_offset & 0xffffffff);
1766                 hashwrite(f, chunk_write, 12);
1767
1768                 chunk_offset += chunks[i].size;
1769         }
1770
1771         if (ctx->report_progress) {
1772                 strbuf_addf(&progress_title,
1773                             Q_("Writing out commit graph in %d pass",
1774                                "Writing out commit graph in %d passes",
1775                                num_chunks),
1776                             num_chunks);
1777                 ctx->progress = start_delayed_progress(
1778                         progress_title.buf,
1779                         num_chunks * ctx->commits.nr);
1780         }
1781
1782         for (i = 0; i < num_chunks; i++) {
1783                 uint64_t start_offset = f->total + f->offset;
1784
1785                 if (chunks[i].write_fn(f, ctx))
1786                         return -1;
1787
1788                 if (f->total + f->offset != start_offset + chunks[i].size)
1789                         BUG("expected to write %"PRId64" bytes to chunk %"PRIx32", but wrote %"PRId64" instead",
1790                             chunks[i].size, chunks[i].id,
1791                             f->total + f->offset - start_offset);
1792         }
1793
1794         stop_progress(&ctx->progress);
1795         strbuf_release(&progress_title);
1796
1797         if (ctx->split && ctx->base_graph_name && ctx->num_commit_graphs_after > 1) {
1798                 char *new_base_hash = xstrdup(oid_to_hex(&ctx->new_base_graph->oid));
1799                 char *new_base_name = get_split_graph_filename(ctx->new_base_graph->odb, new_base_hash);
1800
1801                 free(ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2]);
1802                 free(ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2]);
1803                 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 2] = new_base_name;
1804                 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 2] = new_base_hash;
1805         }
1806
1807         close_commit_graph(ctx->r->objects);
1808         finalize_hashfile(f, file_hash.hash, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1809
1810         if (ctx->split) {
1811                 FILE *chainf = fdopen_lock_file(&lk, "w");
1812                 char *final_graph_name;
1813                 int result;
1814
1815                 close(fd);
1816
1817                 if (!chainf) {
1818                         error(_("unable to open commit-graph chain file"));
1819                         return -1;
1820                 }
1821
1822                 if (ctx->base_graph_name) {
1823                         const char *dest;
1824                         int idx = ctx->num_commit_graphs_after - 1;
1825                         if (ctx->num_commit_graphs_after > 1)
1826                                 idx--;
1827
1828                         dest = ctx->commit_graph_filenames_after[idx];
1829
1830                         if (strcmp(ctx->base_graph_name, dest)) {
1831                                 result = rename(ctx->base_graph_name, dest);
1832
1833                                 if (result) {
1834                                         error(_("failed to rename base commit-graph file"));
1835                                         return -1;
1836                                 }
1837                         }
1838                 } else {
1839                         char *graph_name = get_commit_graph_filename(ctx->odb);
1840                         unlink(graph_name);
1841                 }
1842
1843                 ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1] = xstrdup(oid_to_hex(&file_hash));
1844                 final_graph_name = get_split_graph_filename(ctx->odb,
1845                                         ctx->commit_graph_hash_after[ctx->num_commit_graphs_after - 1]);
1846                 ctx->commit_graph_filenames_after[ctx->num_commit_graphs_after - 1] = final_graph_name;
1847
1848                 result = rename(ctx->graph_name, final_graph_name);
1849
1850                 for (i = 0; i < ctx->num_commit_graphs_after; i++)
1851                         fprintf(lk.tempfile->fp, "%s\n", ctx->commit_graph_hash_after[i]);
1852
1853                 if (result) {
1854                         error(_("failed to rename temporary commit-graph file"));
1855                         return -1;
1856                 }
1857         }
1858
1859         commit_lock_file(&lk);
1860
1861         return 0;
1862 }
1863
1864 static void split_graph_merge_strategy(struct write_commit_graph_context *ctx)
1865 {
1866         struct commit_graph *g;
1867         uint32_t num_commits;
1868         enum commit_graph_split_flags flags = COMMIT_GRAPH_SPLIT_UNSPECIFIED;
1869         uint32_t i;
1870
1871         int max_commits = 0;
1872         int size_mult = 2;
1873
1874         if (ctx->split_opts) {
1875                 max_commits = ctx->split_opts->max_commits;
1876
1877                 if (ctx->split_opts->size_multiple)
1878                         size_mult = ctx->split_opts->size_multiple;
1879
1880                 flags = ctx->split_opts->flags;
1881         }
1882
1883         g = ctx->r->objects->commit_graph;
1884         num_commits = ctx->commits.nr;
1885         if (flags == COMMIT_GRAPH_SPLIT_REPLACE)
1886                 ctx->num_commit_graphs_after = 1;
1887         else
1888                 ctx->num_commit_graphs_after = ctx->num_commit_graphs_before + 1;
1889
1890         if (flags != COMMIT_GRAPH_SPLIT_MERGE_PROHIBITED &&
1891             flags != COMMIT_GRAPH_SPLIT_REPLACE) {
1892                 while (g && (g->num_commits <= size_mult * num_commits ||
1893                             (max_commits && num_commits > max_commits))) {
1894                         if (g->odb != ctx->odb)
1895                                 break;
1896
1897                         num_commits += g->num_commits;
1898                         g = g->base_graph;
1899
1900                         ctx->num_commit_graphs_after--;
1901                 }
1902         }
1903
1904         if (flags != COMMIT_GRAPH_SPLIT_REPLACE)
1905                 ctx->new_base_graph = g;
1906         else if (ctx->num_commit_graphs_after != 1)
1907                 BUG("split_graph_merge_strategy: num_commit_graphs_after "
1908                     "should be 1 with --split=replace");
1909
1910         if (ctx->num_commit_graphs_after == 2) {
1911                 char *old_graph_name = get_commit_graph_filename(g->odb);
1912
1913                 if (!strcmp(g->filename, old_graph_name) &&
1914                     g->odb != ctx->odb) {
1915                         ctx->num_commit_graphs_after = 1;
1916                         ctx->new_base_graph = NULL;
1917                 }
1918
1919                 free(old_graph_name);
1920         }
1921
1922         CALLOC_ARRAY(ctx->commit_graph_filenames_after, ctx->num_commit_graphs_after);
1923         CALLOC_ARRAY(ctx->commit_graph_hash_after, ctx->num_commit_graphs_after);
1924
1925         for (i = 0; i < ctx->num_commit_graphs_after &&
1926                     i < ctx->num_commit_graphs_before; i++)
1927                 ctx->commit_graph_filenames_after[i] = xstrdup(ctx->commit_graph_filenames_before[i]);
1928
1929         i = ctx->num_commit_graphs_before - 1;
1930         g = ctx->r->objects->commit_graph;
1931
1932         while (g) {
1933                 if (i < ctx->num_commit_graphs_after)
1934                         ctx->commit_graph_hash_after[i] = xstrdup(oid_to_hex(&g->oid));
1935
1936                 i--;
1937                 g = g->base_graph;
1938         }
1939 }
1940
1941 static void merge_commit_graph(struct write_commit_graph_context *ctx,
1942                                struct commit_graph *g)
1943 {
1944         uint32_t i;
1945         uint32_t offset = g->num_commits_in_base;
1946
1947         ALLOC_GROW(ctx->commits.list, ctx->commits.nr + g->num_commits, ctx->commits.alloc);
1948
1949         for (i = 0; i < g->num_commits; i++) {
1950                 struct object_id oid;
1951                 struct commit *result;
1952
1953                 display_progress(ctx->progress, i + 1);
1954
1955                 load_oid_from_graph(g, i + offset, &oid);
1956
1957                 /* only add commits if they still exist in the repo */
1958                 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1959
1960                 if (result) {
1961                         ctx->commits.list[ctx->commits.nr] = result;
1962                         ctx->commits.nr++;
1963                 }
1964         }
1965 }
1966
1967 static int commit_compare(const void *_a, const void *_b)
1968 {
1969         const struct commit *a = *(const struct commit **)_a;
1970         const struct commit *b = *(const struct commit **)_b;
1971         return oidcmp(&a->object.oid, &b->object.oid);
1972 }
1973
1974 static void sort_and_scan_merged_commits(struct write_commit_graph_context *ctx)
1975 {
1976         uint32_t i;
1977
1978         if (ctx->report_progress)
1979                 ctx->progress = start_delayed_progress(
1980                                         _("Scanning merged commits"),
1981                                         ctx->commits.nr);
1982
1983         QSORT(ctx->commits.list, ctx->commits.nr, commit_compare);
1984
1985         ctx->num_extra_edges = 0;
1986         for (i = 0; i < ctx->commits.nr; i++) {
1987                 display_progress(ctx->progress, i);
1988
1989                 if (i && oideq(&ctx->commits.list[i - 1]->object.oid,
1990                           &ctx->commits.list[i]->object.oid)) {
1991                         die(_("unexpected duplicate commit id %s"),
1992                             oid_to_hex(&ctx->commits.list[i]->object.oid));
1993                 } else {
1994                         unsigned int num_parents;
1995
1996                         num_parents = commit_list_count(ctx->commits.list[i]->parents);
1997                         if (num_parents > 2)
1998                                 ctx->num_extra_edges += num_parents - 1;
1999                 }
2000         }
2001
2002         stop_progress(&ctx->progress);
2003 }
2004
2005 static void merge_commit_graphs(struct write_commit_graph_context *ctx)
2006 {
2007         struct commit_graph *g = ctx->r->objects->commit_graph;
2008         uint32_t current_graph_number = ctx->num_commit_graphs_before;
2009
2010         while (g && current_graph_number >= ctx->num_commit_graphs_after) {
2011                 current_graph_number--;
2012
2013                 if (ctx->report_progress)
2014                         ctx->progress = start_delayed_progress(_("Merging commit-graph"), 0);
2015
2016                 merge_commit_graph(ctx, g);
2017                 stop_progress(&ctx->progress);
2018
2019                 g = g->base_graph;
2020         }
2021
2022         if (g) {
2023                 ctx->new_base_graph = g;
2024                 ctx->new_num_commits_in_base = g->num_commits + g->num_commits_in_base;
2025         }
2026
2027         if (ctx->new_base_graph)
2028                 ctx->base_graph_name = xstrdup(ctx->new_base_graph->filename);
2029
2030         sort_and_scan_merged_commits(ctx);
2031 }
2032
2033 static void mark_commit_graphs(struct write_commit_graph_context *ctx)
2034 {
2035         uint32_t i;
2036         time_t now = time(NULL);
2037
2038         for (i = ctx->num_commit_graphs_after - 1; i < ctx->num_commit_graphs_before; i++) {
2039                 struct stat st;
2040                 struct utimbuf updated_time;
2041
2042                 stat(ctx->commit_graph_filenames_before[i], &st);
2043
2044                 updated_time.actime = st.st_atime;
2045                 updated_time.modtime = now;
2046                 utime(ctx->commit_graph_filenames_before[i], &updated_time);
2047         }
2048 }
2049
2050 static void expire_commit_graphs(struct write_commit_graph_context *ctx)
2051 {
2052         struct strbuf path = STRBUF_INIT;
2053         DIR *dir;
2054         struct dirent *de;
2055         size_t dirnamelen;
2056         timestamp_t expire_time = time(NULL);
2057
2058         if (ctx->split_opts && ctx->split_opts->expire_time)
2059                 expire_time = ctx->split_opts->expire_time;
2060         if (!ctx->split) {
2061                 char *chain_file_name = get_chain_filename(ctx->odb);
2062                 unlink(chain_file_name);
2063                 free(chain_file_name);
2064                 ctx->num_commit_graphs_after = 0;
2065         }
2066
2067         strbuf_addstr(&path, ctx->odb->path);
2068         strbuf_addstr(&path, "/info/commit-graphs");
2069         dir = opendir(path.buf);
2070
2071         if (!dir)
2072                 goto out;
2073
2074         strbuf_addch(&path, '/');
2075         dirnamelen = path.len;
2076         while ((de = readdir(dir)) != NULL) {
2077                 struct stat st;
2078                 uint32_t i, found = 0;
2079
2080                 strbuf_setlen(&path, dirnamelen);
2081                 strbuf_addstr(&path, de->d_name);
2082
2083                 stat(path.buf, &st);
2084
2085                 if (st.st_mtime > expire_time)
2086                         continue;
2087                 if (path.len < 6 || strcmp(path.buf + path.len - 6, ".graph"))
2088                         continue;
2089
2090                 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2091                         if (!strcmp(ctx->commit_graph_filenames_after[i],
2092                                     path.buf)) {
2093                                 found = 1;
2094                                 break;
2095                         }
2096                 }
2097
2098                 if (!found)
2099                         unlink(path.buf);
2100         }
2101
2102 out:
2103         strbuf_release(&path);
2104 }
2105
2106 int write_commit_graph(struct object_directory *odb,
2107                        struct string_list *pack_indexes,
2108                        struct oidset *commits,
2109                        enum commit_graph_write_flags flags,
2110                        const struct split_commit_graph_opts *split_opts)
2111 {
2112         struct write_commit_graph_context *ctx;
2113         uint32_t i, count_distinct = 0;
2114         int res = 0;
2115         int replace = 0;
2116
2117         if (!commit_graph_compatible(the_repository))
2118                 return 0;
2119
2120         ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
2121         ctx->r = the_repository;
2122         ctx->odb = odb;
2123         ctx->append = flags & COMMIT_GRAPH_WRITE_APPEND ? 1 : 0;
2124         ctx->report_progress = flags & COMMIT_GRAPH_WRITE_PROGRESS ? 1 : 0;
2125         ctx->split = flags & COMMIT_GRAPH_WRITE_SPLIT ? 1 : 0;
2126         ctx->split_opts = split_opts;
2127         ctx->total_bloom_filter_data_size = 0;
2128
2129         if (flags & COMMIT_GRAPH_WRITE_BLOOM_FILTERS)
2130                 ctx->changed_paths = 1;
2131         if (!(flags & COMMIT_GRAPH_NO_WRITE_BLOOM_FILTERS)) {
2132                 struct commit_graph *g;
2133                 prepare_commit_graph_one(ctx->r, ctx->odb);
2134
2135                 g = ctx->r->objects->commit_graph;
2136
2137                 /* We have changed-paths already. Keep them in the next graph */
2138                 if (g && g->chunk_bloom_data) {
2139                         ctx->changed_paths = 1;
2140                         ctx->bloom_settings = g->bloom_filter_settings;
2141                 }
2142         }
2143
2144         if (ctx->split) {
2145                 struct commit_graph *g;
2146                 prepare_commit_graph(ctx->r);
2147
2148                 g = ctx->r->objects->commit_graph;
2149
2150                 while (g) {
2151                         ctx->num_commit_graphs_before++;
2152                         g = g->base_graph;
2153                 }
2154
2155                 if (ctx->num_commit_graphs_before) {
2156                         ALLOC_ARRAY(ctx->commit_graph_filenames_before, ctx->num_commit_graphs_before);
2157                         i = ctx->num_commit_graphs_before;
2158                         g = ctx->r->objects->commit_graph;
2159
2160                         while (g) {
2161                                 ctx->commit_graph_filenames_before[--i] = xstrdup(g->filename);
2162                                 g = g->base_graph;
2163                         }
2164                 }
2165
2166                 if (ctx->split_opts)
2167                         replace = ctx->split_opts->flags & COMMIT_GRAPH_SPLIT_REPLACE;
2168         }
2169
2170         ctx->approx_nr_objects = approximate_object_count();
2171         ctx->oids.alloc = ctx->approx_nr_objects / 32;
2172
2173         if (ctx->split && split_opts && ctx->oids.alloc > split_opts->max_commits)
2174                 ctx->oids.alloc = split_opts->max_commits;
2175
2176         if (ctx->append) {
2177                 prepare_commit_graph_one(ctx->r, ctx->odb);
2178                 if (ctx->r->objects->commit_graph)
2179                         ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
2180         }
2181
2182         if (ctx->oids.alloc < 1024)
2183                 ctx->oids.alloc = 1024;
2184         ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
2185
2186         if (ctx->append && ctx->r->objects->commit_graph) {
2187                 struct commit_graph *g = ctx->r->objects->commit_graph;
2188                 for (i = 0; i < g->num_commits; i++) {
2189                         const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
2190                         hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
2191                 }
2192         }
2193
2194         if (pack_indexes) {
2195                 ctx->order_by_pack = 1;
2196                 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
2197                         goto cleanup;
2198         }
2199
2200         if (commits) {
2201                 if ((res = fill_oids_from_commits(ctx, commits)))
2202                         goto cleanup;
2203         }
2204
2205         if (!pack_indexes && !commits) {
2206                 ctx->order_by_pack = 1;
2207                 fill_oids_from_all_packs(ctx);
2208         }
2209
2210         close_reachable(ctx);
2211
2212         count_distinct = count_distinct_commits(ctx);
2213
2214         if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
2215                 error(_("the commit graph format cannot write %d commits"), count_distinct);
2216                 res = -1;
2217                 goto cleanup;
2218         }
2219
2220         ctx->commits.alloc = count_distinct;
2221         ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
2222
2223         copy_oids_to_commits(ctx);
2224
2225         if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
2226                 error(_("too many commits to write graph"));
2227                 res = -1;
2228                 goto cleanup;
2229         }
2230
2231         if (!ctx->commits.nr && !replace)
2232                 goto cleanup;
2233
2234         if (ctx->split) {
2235                 split_graph_merge_strategy(ctx);
2236
2237                 if (!replace)
2238                         merge_commit_graphs(ctx);
2239         } else
2240                 ctx->num_commit_graphs_after = 1;
2241
2242         compute_generation_numbers(ctx);
2243
2244         if (ctx->changed_paths)
2245                 compute_bloom_filters(ctx);
2246
2247         res = write_commit_graph_file(ctx);
2248
2249         if (ctx->split)
2250                 mark_commit_graphs(ctx);
2251
2252         expire_commit_graphs(ctx);
2253
2254 cleanup:
2255         free(ctx->graph_name);
2256         free(ctx->commits.list);
2257         free(ctx->oids.list);
2258
2259         if (ctx->commit_graph_filenames_after) {
2260                 for (i = 0; i < ctx->num_commit_graphs_after; i++) {
2261                         free(ctx->commit_graph_filenames_after[i]);
2262                         free(ctx->commit_graph_hash_after[i]);
2263                 }
2264
2265                 for (i = 0; i < ctx->num_commit_graphs_before; i++)
2266                         free(ctx->commit_graph_filenames_before[i]);
2267
2268                 free(ctx->commit_graph_filenames_after);
2269                 free(ctx->commit_graph_filenames_before);
2270                 free(ctx->commit_graph_hash_after);
2271         }
2272
2273         free(ctx);
2274
2275         return res;
2276 }
2277
2278 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
2279 static int verify_commit_graph_error;
2280
2281 static void graph_report(const char *fmt, ...)
2282 {
2283         va_list ap;
2284
2285         verify_commit_graph_error = 1;
2286         va_start(ap, fmt);
2287         vfprintf(stderr, fmt, ap);
2288         fprintf(stderr, "\n");
2289         va_end(ap);
2290 }
2291
2292 #define GENERATION_ZERO_EXISTS 1
2293 #define GENERATION_NUMBER_EXISTS 2
2294
2295 int verify_commit_graph(struct repository *r, struct commit_graph *g, int flags)
2296 {
2297         uint32_t i, cur_fanout_pos = 0;
2298         struct object_id prev_oid, cur_oid, checksum;
2299         int generation_zero = 0;
2300         struct hashfile *f;
2301         int devnull;
2302         struct progress *progress = NULL;
2303         int local_error = 0;
2304
2305         if (!g) {
2306                 graph_report("no commit-graph file loaded");
2307                 return 1;
2308         }
2309
2310         verify_commit_graph_error = verify_commit_graph_lite(g);
2311         if (verify_commit_graph_error)
2312                 return verify_commit_graph_error;
2313
2314         devnull = open("/dev/null", O_WRONLY);
2315         f = hashfd(devnull, NULL);
2316         hashwrite(f, g->data, g->data_len - g->hash_len);
2317         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
2318         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
2319                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
2320                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
2321         }
2322
2323         for (i = 0; i < g->num_commits; i++) {
2324                 struct commit *graph_commit;
2325
2326                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2327
2328                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
2329                         graph_report(_("commit-graph has incorrect OID order: %s then %s"),
2330                                      oid_to_hex(&prev_oid),
2331                                      oid_to_hex(&cur_oid));
2332
2333                 oidcpy(&prev_oid, &cur_oid);
2334
2335                 while (cur_oid.hash[0] > cur_fanout_pos) {
2336                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2337
2338                         if (i != fanout_value)
2339                                 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2340                                              cur_fanout_pos, fanout_value, i);
2341                         cur_fanout_pos++;
2342                 }
2343
2344                 graph_commit = lookup_commit(r, &cur_oid);
2345                 if (!parse_commit_in_graph_one(r, g, graph_commit))
2346                         graph_report(_("failed to parse commit %s from commit-graph"),
2347                                      oid_to_hex(&cur_oid));
2348         }
2349
2350         while (cur_fanout_pos < 256) {
2351                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
2352
2353                 if (g->num_commits != fanout_value)
2354                         graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
2355                                      cur_fanout_pos, fanout_value, i);
2356
2357                 cur_fanout_pos++;
2358         }
2359
2360         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
2361                 return verify_commit_graph_error;
2362
2363         if (flags & COMMIT_GRAPH_WRITE_PROGRESS)
2364                 progress = start_progress(_("Verifying commits in commit graph"),
2365                                         g->num_commits);
2366
2367         for (i = 0; i < g->num_commits; i++) {
2368                 struct commit *graph_commit, *odb_commit;
2369                 struct commit_list *graph_parents, *odb_parents;
2370                 uint32_t max_generation = 0;
2371                 uint32_t generation;
2372
2373                 display_progress(progress, i + 1);
2374                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
2375
2376                 graph_commit = lookup_commit(r, &cur_oid);
2377                 odb_commit = (struct commit *)create_object(r, &cur_oid, alloc_commit_node(r));
2378                 if (parse_commit_internal(odb_commit, 0, 0)) {
2379                         graph_report(_("failed to parse commit %s from object database for commit-graph"),
2380                                      oid_to_hex(&cur_oid));
2381                         continue;
2382                 }
2383
2384                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
2385                            get_commit_tree_oid(odb_commit)))
2386                         graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
2387                                      oid_to_hex(&cur_oid),
2388                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
2389                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
2390
2391                 graph_parents = graph_commit->parents;
2392                 odb_parents = odb_commit->parents;
2393
2394                 while (graph_parents) {
2395                         if (odb_parents == NULL) {
2396                                 graph_report(_("commit-graph parent list for commit %s is too long"),
2397                                              oid_to_hex(&cur_oid));
2398                                 break;
2399                         }
2400
2401                         /* parse parent in case it is in a base graph */
2402                         parse_commit_in_graph_one(r, g, graph_parents->item);
2403
2404                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
2405                                 graph_report(_("commit-graph parent for %s is %s != %s"),
2406                                              oid_to_hex(&cur_oid),
2407                                              oid_to_hex(&graph_parents->item->object.oid),
2408                                              oid_to_hex(&odb_parents->item->object.oid));
2409
2410                         generation = commit_graph_generation(graph_parents->item);
2411                         if (generation > max_generation)
2412                                 max_generation = generation;
2413
2414                         graph_parents = graph_parents->next;
2415                         odb_parents = odb_parents->next;
2416                 }
2417
2418                 if (odb_parents != NULL)
2419                         graph_report(_("commit-graph parent list for commit %s terminates early"),
2420                                      oid_to_hex(&cur_oid));
2421
2422                 if (!commit_graph_generation(graph_commit)) {
2423                         if (generation_zero == GENERATION_NUMBER_EXISTS)
2424                                 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
2425                                              oid_to_hex(&cur_oid));
2426                         generation_zero = GENERATION_ZERO_EXISTS;
2427                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
2428                         graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
2429                                      oid_to_hex(&cur_oid));
2430
2431                 if (generation_zero == GENERATION_ZERO_EXISTS)
2432                         continue;
2433
2434                 /*
2435                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
2436                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
2437                  * extra logic in the following condition.
2438                  */
2439                 if (max_generation == GENERATION_NUMBER_MAX)
2440                         max_generation--;
2441
2442                 generation = commit_graph_generation(graph_commit);
2443                 if (generation != max_generation + 1)
2444                         graph_report(_("commit-graph generation for commit %s is %u != %u"),
2445                                      oid_to_hex(&cur_oid),
2446                                      generation,
2447                                      max_generation + 1);
2448
2449                 if (graph_commit->date != odb_commit->date)
2450                         graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
2451                                      oid_to_hex(&cur_oid),
2452                                      graph_commit->date,
2453                                      odb_commit->date);
2454         }
2455         stop_progress(&progress);
2456
2457         local_error = verify_commit_graph_error;
2458
2459         if (!(flags & COMMIT_GRAPH_VERIFY_SHALLOW) && g->base_graph)
2460                 local_error |= verify_commit_graph(r, g->base_graph, flags);
2461
2462         return local_error;
2463 }
2464
2465 void free_commit_graph(struct commit_graph *g)
2466 {
2467         if (!g)
2468                 return;
2469         if (g->data) {
2470                 munmap((void *)g->data, g->data_len);
2471                 g->data = NULL;
2472         }
2473         free(g->filename);
2474         free(g->bloom_filter_settings);
2475         free(g);
2476 }
2477
2478 void disable_commit_graph(struct repository *r)
2479 {
2480         r->commit_graph_disabled = 1;
2481 }