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