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