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

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