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         return xstrfmt("%s/info/commit-graph", obj_dir);
  47 }
  48
  49 static char *get_split_graph_filename(const char *obj_dir,
  50                                       const char *oid_hex)
  51 {
  52         return xstrfmt("%s/info/commit-graphs/graph-%s.graph",
  53                        obj_dir,
  54                        oid_hex);
  55 }
  56
  57 static char *get_chain_filename(const char *obj_dir)
  58 {
  59         return xstrfmt("%s/info/commit-graphs/commit-graph-chain", obj_dir);
  60 }
  61
  62 static uint8_t oid_version(void)
  63 {
  64         return 1;
  65 }
  66
  67 static struct commit_graph *alloc_commit_graph(void)
  68 {
  69         struct commit_graph *g = xcalloc(1, sizeof(*g));
  70         g->graph_fd = -1;
  71
  72         return g;
  73 }
  74
  75 extern int read_replace_refs;
  76
  77 static int commit_graph_compatible(struct repository *r)
  78 {
  79         if (!r->gitdir)
  80                 return 0;
  81
  82         if (read_replace_refs) {
  83                 prepare_replace_object(r);
  84                 if (hashmap_get_size(&r->objects->replace_map->map))
  85                         return 0;
  86         }
  87
  88         prepare_commit_graft(r);
  89         if (r->parsed_objects && r->parsed_objects->grafts_nr)
  90                 return 0;
  91         if (is_repository_shallow(r))
  92                 return 0;
  93
  94         return 1;
  95 }
  96
  97 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
  98 {
  99         *fd = git_open(graph_file);
 100         if (*fd < 0)
 101                 return 0;
 102         if (fstat(*fd, st)) {
 103                 close(*fd);
 104                 return 0;
 105         }
 106         return 1;
 107 }
 108
 109 struct commit_graph *load_commit_graph_one_fd_st(int fd, struct stat *st)
 110 {
 111         void *graph_map;
 112         size_t graph_size;
 113         struct commit_graph *ret;
 114
 115         graph_size = xsize_t(st->st_size);
 116
 117         if (graph_size < GRAPH_MIN_SIZE) {
 118                 close(fd);
 119                 error(_("commit-graph file is too small"));
 120                 return NULL;
 121         }
 122         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
 123         ret = parse_commit_graph(graph_map, fd, graph_size);
 124
 125         if (!ret) {
 126                 munmap(graph_map, graph_size);
 127                 close(fd);
 128         }
 129
 130         return ret;
 131 }
 132
 133 static int verify_commit_graph_lite(struct commit_graph *g)
 134 {
 135         /*
 136          * Basic validation shared between parse_commit_graph()
 137          * which'll be called every time the graph is used, and the
 138          * much more expensive verify_commit_graph() used by
 139          * "commit-graph verify".
 140          *
 141          * There should only be very basic checks here to ensure that
 142          * we don't e.g. segfault in fill_commit_in_graph(), but
 143          * because this is a very hot codepath nothing that e.g. loops
 144          * over g->num_commits, or runs a checksum on the commit-graph
 145          * itself.
 146          */
 147         if (!g->chunk_oid_fanout) {
 148                 error("commit-graph is missing the OID Fanout chunk");
 149                 return 1;
 150         }
 151         if (!g->chunk_oid_lookup) {
 152                 error("commit-graph is missing the OID Lookup chunk");
 153                 return 1;
 154         }
 155         if (!g->chunk_commit_data) {
 156                 error("commit-graph is missing the Commit Data chunk");
 157                 return 1;
 158         }
 159
 160         return 0;
 161 }
 162
 163 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
 164                                         size_t graph_size)
 165 {
 166         const unsigned char *data, *chunk_lookup;
 167         uint32_t i;
 168         struct commit_graph *graph;
 169         uint64_t last_chunk_offset;
 170         uint32_t last_chunk_id;
 171         uint32_t graph_signature;
 172         unsigned char graph_version, hash_version;
 173
 174         if (!graph_map)
 175                 return NULL;
 176
 177         if (graph_size < GRAPH_MIN_SIZE)
 178                 return NULL;
 179
 180         data = (const unsigned char *)graph_map;
 181
 182         graph_signature = get_be32(data);
 183         if (graph_signature != GRAPH_SIGNATURE) {
 184                 error(_("commit-graph signature %X does not match signature %X"),
 185                       graph_signature, GRAPH_SIGNATURE);
 186                 return NULL;
 187         }
 188
 189         graph_version = *(unsigned char*)(data + 4);
 190         if (graph_version != GRAPH_VERSION) {
 191                 error(_("commit-graph version %X does not match version %X"),
 192                       graph_version, GRAPH_VERSION);
 193                 return NULL;
 194         }
 195
 196         hash_version = *(unsigned char*)(data + 5);
 197         if (hash_version != oid_version()) {
 198                 error(_("commit-graph hash version %X does not match version %X"),
 199                       hash_version, oid_version());
 200                 return NULL;
 201         }
 202
 203         graph = alloc_commit_graph();
 204
 205         graph->hash_len = the_hash_algo->rawsz;
 206         graph->num_chunks = *(unsigned char*)(data + 6);
 207         graph->graph_fd = fd;
 208         graph->data = graph_map;
 209         graph->data_len = graph_size;
 210
 211         last_chunk_id = 0;
 212         last_chunk_offset = 8;
 213         chunk_lookup = data + 8;
 214         for (i = 0; i < graph->num_chunks; i++) {
 215                 uint32_t chunk_id;
 216                 uint64_t chunk_offset;
 217                 int chunk_repeated = 0;
 218
 219                 if (data + graph_size - chunk_lookup <
 220                     GRAPH_CHUNKLOOKUP_WIDTH) {
 221                         error(_("commit-graph chunk lookup table entry missing; file may be incomplete"));
 222                         free(graph);
 223                         return NULL;
 224                 }
 225
 226                 chunk_id = get_be32(chunk_lookup + 0);
 227                 chunk_offset = get_be64(chunk_lookup + 4);
 228
 229                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 230
 231                 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
 232                         error(_("commit-graph improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 233                               (uint32_t)chunk_offset);
 234                         free(graph);
 235                         return NULL;
 236                 }
 237
 238                 switch (chunk_id) {
 239                 case GRAPH_CHUNKID_OIDFANOUT:
 240                         if (graph->chunk_oid_fanout)
 241                                 chunk_repeated = 1;
 242                         else
 243                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 244                         break;
 245
 246                 case GRAPH_CHUNKID_OIDLOOKUP:
 247                         if (graph->chunk_oid_lookup)
 248                                 chunk_repeated = 1;
 249                         else
 250                                 graph->chunk_oid_lookup = data + chunk_offset;
 251                         break;
 252
 253                 case GRAPH_CHUNKID_DATA:
 254                         if (graph->chunk_commit_data)
 255                                 chunk_repeated = 1;
 256                         else
 257                                 graph->chunk_commit_data = data + chunk_offset;
 258                         break;
 259
 260                 case GRAPH_CHUNKID_EXTRAEDGES:
 261                         if (graph->chunk_extra_edges)
 262                                 chunk_repeated = 1;
 263                         else
 264                                 graph->chunk_extra_edges = data + chunk_offset;
 265                         break;
 266
 267                 case GRAPH_CHUNKID_BASE:
 268                         if (graph->chunk_base_graphs)
 269                                 chunk_repeated = 1;
 270                         else
 271                                 graph->chunk_base_graphs = data + chunk_offset;
 272                 }
 273
 274                 if (chunk_repeated) {
 275                         error(_("commit-graph chunk id %08x appears multiple times"), chunk_id);
 276                         free(graph);
 277                         return NULL;
 278                 }
 279
 280                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 281                 {
 282                         graph->num_commits = (chunk_offset - last_chunk_offset)
 283                                              / graph->hash_len;
 284                 }
 285
 286                 last_chunk_id = chunk_id;
 287                 last_chunk_offset = chunk_offset;
 288         }
 289
 290         hashcpy(graph->oid.hash, graph->data + graph->data_len - graph->hash_len);
 291
 292         if (verify_commit_graph_lite(graph))
 293                 return NULL;
 294
 295         return graph;
 296 }
 297
 298 static struct commit_graph *load_commit_graph_one(const char *graph_file)
 299 {
 300
 301         struct stat st;
 302         int fd;
 303         int open_ok = open_commit_graph(graph_file, &fd, &st);
 304
 305         if (!open_ok)
 306                 return NULL;
 307
 308         return load_commit_graph_one_fd_st(fd, &st);
 309 }
 310
 311 static struct commit_graph *load_commit_graph_v1(struct repository *r, const char *obj_dir)
 312 {
 313         char *graph_name = get_commit_graph_filename(obj_dir);
 314         struct commit_graph *g = load_commit_graph_one(graph_name);
 315         free(graph_name);
 316
 317         return g;
 318 }
 319
 320 static int add_graph_to_chain(struct commit_graph *g,
 321                               struct commit_graph *chain,
 322                               struct object_id *oids,
 323                               int n)
 324 {
 325         struct commit_graph *cur_g = chain;
 326
 327         if (n && !g->chunk_base_graphs) {
 328                 warning(_("commit-graph has no base graphs chunk"));
 329                 return 0;
 330         }
 331
 332         while (n) {
 333                 n--;
 334
 335                 if (!cur_g ||
 336                     !oideq(&oids[n], &cur_g->oid) ||
 337                     !hasheq(oids[n].hash, g->chunk_base_graphs + g->hash_len * n)) {
 338                         warning(_("commit-graph chain does not match"));
 339                         return 0;
 340                 }
 341
 342                 cur_g = cur_g->base_graph;
 343         }
 344
 345         g->base_graph = chain;
 346
 347         if (chain)
 348                 g->num_commits_in_base = chain->num_commits + chain->num_commits_in_base;
 349
 350         return 1;
 351 }
 352
 353 static struct commit_graph *load_commit_graph_chain(struct repository *r, const char *obj_dir)
 354 {
 355         struct commit_graph *graph_chain = NULL;
 356         struct strbuf line = STRBUF_INIT;
 357         struct stat st;
 358         struct object_id *oids;
 359         int i = 0, valid = 1, count;
 360         char *chain_name = get_chain_filename(obj_dir);
 361         FILE *fp;
 362         int stat_res;
 363
 364         fp = fopen(chain_name, "r");
 365         stat_res = stat(chain_name, &st);
 366         free(chain_name);
 367
 368         if (!fp ||
 369             stat_res ||
 370             st.st_size <= the_hash_algo->hexsz)
 371                 return NULL;
 372
 373         count = st.st_size / (the_hash_algo->hexsz + 1);
 374         oids = xcalloc(count, sizeof(struct object_id));
 375
 376         for (i = 0; i < count && valid; i++) {
 377                 char *graph_name;
 378                 struct commit_graph *g;
 379
 380                 if (strbuf_getline_lf(&line, fp) == EOF)
 381                         break;
 382
 383                 if (get_oid_hex(line.buf, &oids[i])) {
 384                         warning(_("invalid commit-graph chain: line '%s' not a hash"),
 385                                 line.buf);
 386                         valid = 0;
 387                         break;
 388                 }
 389
 390                 graph_name = get_split_graph_filename(obj_dir, line.buf);
 391                 g = load_commit_graph_one(graph_name);
 392                 free(graph_name);
 393
 394                 if (g && add_graph_to_chain(g, graph_chain, oids, i))
 395                         graph_chain = g;
 396                 else
 397                         valid = 0;
 398         }
 399
 400         free(oids);
 401         fclose(fp);
 402
 403         return graph_chain;
 404 }
 405
 406 static struct commit_graph *read_commit_graph_one(struct repository *r, const char *obj_dir)
 407 {
 408         struct commit_graph *g = load_commit_graph_v1(r, obj_dir);
 409
 410         if (!g)
 411                 g = load_commit_graph_chain(r, obj_dir);
 412
 413         return g;
 414 }
 415
 416 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 417 {
 418
 419         if (r->objects->commit_graph)
 420                 return;
 421
 422         r->objects->commit_graph = read_commit_graph_one(r, obj_dir);
 423 }
 424
 425 /*
 426  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 427  *
 428  * On the first invocation, this function attemps to load the commit
 429  * graph if the_repository is configured to have one.
 430  */
 431 static int prepare_commit_graph(struct repository *r)
 432 {
 433         struct object_directory *odb;
 434         int config_value;
 435
 436         if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
 437                 die("dying as requested by the '%s' variable on commit-graph load!",
 438                     GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD);
 439
 440         if (r->objects->commit_graph_attempted)
 441                 return !!r->objects->commit_graph;
 442         r->objects->commit_graph_attempted = 1;
 443
 444         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 445             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 446             !config_value))
 447                 /*
 448                  * This repository is not configured to use commit graphs, so
 449                  * do not load one. (But report commit_graph_attempted anyway
 450                  * so that commit graph loading is not attempted again for this
 451                  * repository.)
 452                  */
 453                 return 0;
 454
 455         if (!commit_graph_compatible(r))
 456                 return 0;
 457
 458         prepare_alt_odb(r);
 459         for (odb = r->objects->odb;
 460              !r->objects->commit_graph && odb;
 461              odb = odb->next)
 462                 prepare_commit_graph_one(r, odb->path);
 463         return !!r->objects->commit_graph;
 464 }
 465
 466 int generation_numbers_enabled(struct repository *r)
 467 {
 468         uint32_t first_generation;
 469         struct commit_graph *g;
 470         if (!prepare_commit_graph(r))
 471                return 0;
 472
 473         g = r->objects->commit_graph;
 474
 475         if (!g->num_commits)
 476                 return 0;
 477
 478         first_generation = get_be32(g->chunk_commit_data +
 479                                     g->hash_len + 8) >> 2;
 480
 481         return !!first_generation;
 482 }
 483
 484 static void close_commit_graph_one(struct commit_graph *g)
 485 {
 486         if (!g)
 487                 return;
 488
 489         close_commit_graph_one(g->base_graph);
 490         free_commit_graph(g);
 491 }
 492
 493 void close_commit_graph(struct raw_object_store *o)
 494 {
 495         close_commit_graph_one(o->commit_graph);
 496         o->commit_graph = NULL;
 497 }
 498
 499 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 500 {
 501         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 502                             g->chunk_oid_lookup, g->hash_len, pos);
 503 }
 504
 505 static void load_oid_from_graph(struct commit_graph *g,
 506                                 uint32_t pos,
 507                                 struct object_id *oid)
 508 {
 509         uint32_t lex_index;
 510
 511         while (g && pos < g->num_commits_in_base)
 512                 g = g->base_graph;
 513
 514         if (!g)
 515                 BUG("NULL commit-graph");
 516
 517         if (pos >= g->num_commits + g->num_commits_in_base)
 518                 die(_("invalid commit position. commit-graph is likely corrupt"));
 519
 520         lex_index = pos - g->num_commits_in_base;
 521
 522         hashcpy(oid->hash, g->chunk_oid_lookup + g->hash_len * lex_index);
 523 }
 524
 525 static struct commit_list **insert_parent_or_die(struct repository *r,
 526                                                  struct commit_graph *g,
 527                                                  uint32_t pos,
 528                                                  struct commit_list **pptr)
 529 {
 530         struct commit *c;
 531         struct object_id oid;
 532
 533         if (pos >= g->num_commits + g->num_commits_in_base)
 534                 die("invalid parent position %"PRIu32, pos);
 535
 536         load_oid_from_graph(g, pos, &oid);
 537         c = lookup_commit(r, &oid);
 538         if (!c)
 539                 die(_("could not find commit %s"), oid_to_hex(&oid));
 540         c->graph_pos = pos;
 541         return &commit_list_insert(c, pptr)->next;
 542 }
 543
 544 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 545 {
 546         const unsigned char *commit_data;
 547         uint32_t lex_index;
 548
 549         while (pos < g->num_commits_in_base)
 550                 g = g->base_graph;
 551
 552         lex_index = pos - g->num_commits_in_base;
 553         commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * lex_index;
 554         item->graph_pos = pos;
 555         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 556 }
 557
 558 static int fill_commit_in_graph(struct repository *r,
 559                                 struct commit *item,
 560                                 struct commit_graph *g, uint32_t pos)
 561 {
 562         uint32_t edge_value;
 563         uint32_t *parent_data_ptr;
 564         uint64_t date_low, date_high;
 565         struct commit_list **pptr;
 566         const unsigned char *commit_data;
 567         uint32_t lex_index;
 568
 569         while (pos < g->num_commits_in_base)
 570                 g = g->base_graph;
 571
 572         if (pos >= g->num_commits + g->num_commits_in_base)
 573                 die(_("invalid commit position. commit-graph is likely corrupt"));
 574
 575         /*
 576          * Store the "full" position, but then use the
 577          * "local" position for the rest of the calculation.
 578          */
 579         item->graph_pos = pos;
 580         lex_index = pos - g->num_commits_in_base;
 581
 582         commit_data = g->chunk_commit_data + (g->hash_len + 16) * lex_index;
 583
 584         item->object.parsed = 1;
 585
 586         item->maybe_tree = NULL;
 587
 588         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 589         date_low = get_be32(commit_data + g->hash_len + 12);
 590         item->date = (timestamp_t)((date_high << 32) | date_low);
 591
 592         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 593
 594         pptr = &item->parents;
 595
 596         edge_value = get_be32(commit_data + g->hash_len);
 597         if (edge_value == GRAPH_PARENT_NONE)
 598                 return 1;
 599         pptr = insert_parent_or_die(r, g, edge_value, pptr);
 600
 601         edge_value = get_be32(commit_data + g->hash_len + 4);
 602         if (edge_value == GRAPH_PARENT_NONE)
 603                 return 1;
 604         if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
 605                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
 606                 return 1;
 607         }
 608
 609         parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
 610                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 611         do {
 612                 edge_value = get_be32(parent_data_ptr);
 613                 pptr = insert_parent_or_die(r, g,
 614                                             edge_value & GRAPH_EDGE_LAST_MASK,
 615                                             pptr);
 616                 parent_data_ptr++;
 617         } while (!(edge_value & GRAPH_LAST_EDGE));
 618
 619         return 1;
 620 }
 621
 622 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 623 {
 624         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 625                 *pos = item->graph_pos;
 626                 return 1;
 627         } else {
 628                 struct commit_graph *cur_g = g;
 629                 uint32_t lex_index;
 630
 631                 while (cur_g && !bsearch_graph(cur_g, &(item->object.oid), &lex_index))
 632                         cur_g = cur_g->base_graph;
 633
 634                 if (cur_g) {
 635                         *pos = lex_index + cur_g->num_commits_in_base;
 636                         return 1;
 637                 }
 638
 639                 return 0;
 640         }
 641 }
 642
 643 static int parse_commit_in_graph_one(struct repository *r,
 644                                      struct commit_graph *g,
 645                                      struct commit *item)
 646 {
 647         uint32_t pos;
 648
 649         if (item->object.parsed)
 650                 return 1;
 651
 652         if (find_commit_in_graph(item, g, &pos))
 653                 return fill_commit_in_graph(r, item, g, pos);
 654
 655         return 0;
 656 }
 657
 658 int parse_commit_in_graph(struct repository *r, struct commit *item)
 659 {
 660         if (!prepare_commit_graph(r))
 661                 return 0;
 662         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
 663 }
 664
 665 void load_commit_graph_info(struct repository *r, struct commit *item)
 666 {
 667         uint32_t pos;
 668         if (!prepare_commit_graph(r))
 669                 return;
 670         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 671                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 672 }
 673
 674 static struct tree *load_tree_for_commit(struct repository *r,
 675                                          struct commit_graph *g,
 676                                          struct commit *c)
 677 {
 678         struct object_id oid;
 679         const unsigned char *commit_data;
 680
 681         while (c->graph_pos < g->num_commits_in_base)
 682                 g = g->base_graph;
 683
 684         commit_data = g->chunk_commit_data +
 685                         GRAPH_DATA_WIDTH * (c->graph_pos - g->num_commits_in_base);
 686
 687         hashcpy(oid.hash, commit_data);
 688         c->maybe_tree = lookup_tree(r, &oid);
 689
 690         return c->maybe_tree;
 691 }
 692
 693 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
 694                                                  struct commit_graph *g,
 695                                                  const struct commit *c)
 696 {
 697         if (c->maybe_tree)
 698                 return c->maybe_tree;
 699         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 700                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 701
 702         return load_tree_for_commit(r, g, (struct commit *)c);
 703 }
 704
 705 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 706 {
 707         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
 708 }
 709
 710 struct packed_commit_list {
 711         struct commit **list;
 712         int nr;
 713         int alloc;
 714 };
 715
 716 struct packed_oid_list {
 717         struct object_id *list;
 718         int nr;
 719         int alloc;
 720 };
 721
 722 struct write_commit_graph_context {
 723         struct repository *r;
 724         const char *obj_dir;
 725         char *graph_name;
 726         struct packed_oid_list oids;
 727         struct packed_commit_list commits;
 728         int num_extra_edges;
 729         unsigned long approx_nr_objects;
 730         struct progress *progress;
 731         int progress_done;
 732         uint64_t progress_cnt;
 733         unsigned append:1,
 734                  report_progress:1;
 735 };
 736
 737 static void write_graph_chunk_fanout(struct hashfile *f,
 738                                      struct write_commit_graph_context *ctx)
 739 {
 740         int i, count = 0;
 741         struct commit **list = ctx->commits.list;
 742
 743         /*
 744          * Write the first-level table (the list is sorted,
 745          * but we use a 256-entry lookup to be able to avoid
 746          * having to do eight extra binary search iterations).
 747          */
 748         for (i = 0; i < 256; i++) {
 749                 while (count < ctx->commits.nr) {
 750                         if ((*list)->object.oid.hash[0] != i)
 751                                 break;
 752                         display_progress(ctx->progress, ++ctx->progress_cnt);
 753                         count++;
 754                         list++;
 755                 }
 756
 757                 hashwrite_be32(f, count);
 758         }
 759 }
 760
 761 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 762                                    struct write_commit_graph_context *ctx)
 763 {
 764         struct commit **list = ctx->commits.list;
 765         int count;
 766         for (count = 0; count < ctx->commits.nr; count++, list++) {
 767                 display_progress(ctx->progress, ++ctx->progress_cnt);
 768                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 769         }
 770 }
 771
 772 static const unsigned char *commit_to_sha1(size_t index, void *table)
 773 {
 774         struct commit **commits = table;
 775         return commits[index]->object.oid.hash;
 776 }
 777
 778 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 779                                    struct write_commit_graph_context *ctx)
 780 {
 781         struct commit **list = ctx->commits.list;
 782         struct commit **last = ctx->commits.list + ctx->commits.nr;
 783         uint32_t num_extra_edges = 0;
 784
 785         while (list < last) {
 786                 struct commit_list *parent;
 787                 int edge_value;
 788                 uint32_t packedDate[2];
 789                 display_progress(ctx->progress, ++ctx->progress_cnt);
 790
 791                 parse_commit_no_graph(*list);
 792                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 793
 794                 parent = (*list)->parents;
 795
 796                 if (!parent)
 797                         edge_value = GRAPH_PARENT_NONE;
 798                 else {
 799                         edge_value = sha1_pos(parent->item->object.oid.hash,
 800                                               ctx->commits.list,
 801                                               ctx->commits.nr,
 802                                               commit_to_sha1);
 803
 804                         if (edge_value < 0)
 805                                 BUG("missing parent %s for commit %s",
 806                                     oid_to_hex(&parent->item->object.oid),
 807                                     oid_to_hex(&(*list)->object.oid));
 808                 }
 809
 810                 hashwrite_be32(f, edge_value);
 811
 812                 if (parent)
 813                         parent = parent->next;
 814
 815                 if (!parent)
 816                         edge_value = GRAPH_PARENT_NONE;
 817                 else if (parent->next)
 818                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
 819                 else {
 820                         edge_value = sha1_pos(parent->item->object.oid.hash,
 821                                               ctx->commits.list,
 822                                               ctx->commits.nr,
 823                                               commit_to_sha1);
 824                         if (edge_value < 0)
 825                                 BUG("missing parent %s for commit %s",
 826                                     oid_to_hex(&parent->item->object.oid),
 827                                     oid_to_hex(&(*list)->object.oid));
 828                 }
 829
 830                 hashwrite_be32(f, edge_value);
 831
 832                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
 833                         do {
 834                                 num_extra_edges++;
 835                                 parent = parent->next;
 836                         } while (parent);
 837                 }
 838
 839                 if (sizeof((*list)->date) > 4)
 840                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 841                 else
 842                         packedDate[0] = 0;
 843
 844                 packedDate[0] |= htonl((*list)->generation << 2);
 845
 846                 packedDate[1] = htonl((*list)->date);
 847                 hashwrite(f, packedDate, 8);
 848
 849                 list++;
 850         }
 851 }
 852
 853 static void write_graph_chunk_extra_edges(struct hashfile *f,
 854                                           struct write_commit_graph_context *ctx)
 855 {
 856         struct commit **list = ctx->commits.list;
 857         struct commit **last = ctx->commits.list + ctx->commits.nr;
 858         struct commit_list *parent;
 859
 860         while (list < last) {
 861                 int num_parents = 0;
 862
 863                 display_progress(ctx->progress, ++ctx->progress_cnt);
 864
 865                 for (parent = (*list)->parents; num_parents < 3 && parent;
 866                      parent = parent->next)
 867                         num_parents++;
 868
 869                 if (num_parents <= 2) {
 870                         list++;
 871                         continue;
 872                 }
 873
 874                 /* Since num_parents > 2, this initializer is safe. */
 875                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 876                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 877                                                   ctx->commits.list,
 878                                                   ctx->commits.nr,
 879                                                   commit_to_sha1);
 880
 881                         if (edge_value < 0)
 882                                 BUG("missing parent %s for commit %s",
 883                                     oid_to_hex(&parent->item->object.oid),
 884                                     oid_to_hex(&(*list)->object.oid));
 885                         else if (!parent->next)
 886                                 edge_value |= GRAPH_LAST_EDGE;
 887
 888                         hashwrite_be32(f, edge_value);
 889                 }
 890
 891                 list++;
 892         }
 893 }
 894
 895 static int oid_compare(const void *_a, const void *_b)
 896 {
 897         const struct object_id *a = (const struct object_id *)_a;
 898         const struct object_id *b = (const struct object_id *)_b;
 899         return oidcmp(a, b);
 900 }
 901
 902 static int add_packed_commits(const struct object_id *oid,
 903                               struct packed_git *pack,
 904                               uint32_t pos,
 905                               void *data)
 906 {
 907         struct write_commit_graph_context *ctx = (struct write_commit_graph_context*)data;
 908         enum object_type type;
 909         off_t offset = nth_packed_object_offset(pack, pos);
 910         struct object_info oi = OBJECT_INFO_INIT;
 911
 912         if (ctx->progress)
 913                 display_progress(ctx->progress, ++ctx->progress_done);
 914
 915         oi.typep = &type;
 916         if (packed_object_info(ctx->r, pack, offset, &oi) < 0)
 917                 die(_("unable to get type of object %s"), oid_to_hex(oid));
 918
 919         if (type != OBJ_COMMIT)
 920                 return 0;
 921
 922         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
 923         oidcpy(&(ctx->oids.list[ctx->oids.nr]), oid);
 924         ctx->oids.nr++;
 925
 926         return 0;
 927 }
 928
 929 static void add_missing_parents(struct write_commit_graph_context *ctx, struct commit *commit)
 930 {
 931         struct commit_list *parent;
 932         for (parent = commit->parents; parent; parent = parent->next) {
 933                 if (!(parent->item->object.flags & UNINTERESTING)) {
 934                         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
 935                         oidcpy(&ctx->oids.list[ctx->oids.nr], &(parent->item->object.oid));
 936                         ctx->oids.nr++;
 937                         parent->item->object.flags |= UNINTERESTING;
 938                 }
 939         }
 940 }
 941
 942 static void close_reachable(struct write_commit_graph_context *ctx)
 943 {
 944         int i;
 945         struct commit *commit;
 946
 947         if (ctx->report_progress)
 948                 ctx->progress = start_delayed_progress(
 949                                         _("Loading known commits in commit graph"),
 950                                         ctx->oids.nr);
 951         for (i = 0; i < ctx->oids.nr; i++) {
 952                 display_progress(ctx->progress, i + 1);
 953                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
 954                 if (commit)
 955                         commit->object.flags |= UNINTERESTING;
 956         }
 957         stop_progress(&ctx->progress);
 958
 959         /*
 960          * As this loop runs, ctx->oids.nr may grow, but not more
 961          * than the number of missing commits in the reachable
 962          * closure.
 963          */
 964         if (ctx->report_progress)
 965                 ctx->progress = start_delayed_progress(
 966                                         _("Expanding reachable commits in commit graph"),
 967                                         ctx->oids.nr);
 968         for (i = 0; i < ctx->oids.nr; i++) {
 969                 display_progress(ctx->progress, i + 1);
 970                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
 971
 972                 if (commit && !parse_commit_no_graph(commit))
 973                         add_missing_parents(ctx, commit);
 974         }
 975         stop_progress(&ctx->progress);
 976
 977         if (ctx->report_progress)
 978                 ctx->progress = start_delayed_progress(
 979                                         _("Clearing commit marks in commit graph"),
 980                                         ctx->oids.nr);
 981         for (i = 0; i < ctx->oids.nr; i++) {
 982                 display_progress(ctx->progress, i + 1);
 983                 commit = lookup_commit(ctx->r, &ctx->oids.list[i]);
 984
 985                 if (commit)
 986                         commit->object.flags &= ~UNINTERESTING;
 987         }
 988         stop_progress(&ctx->progress);
 989 }
 990
 991 static void compute_generation_numbers(struct write_commit_graph_context *ctx)
 992 {
 993         int i;
 994         struct commit_list *list = NULL;
 995
 996         if (ctx->report_progress)
 997                 ctx->progress = start_progress(
 998                                         _("Computing commit graph generation numbers"),
 999                                         ctx->commits.nr);
1000         for (i = 0; i < ctx->commits.nr; i++) {
1001                 display_progress(ctx->progress, i + 1);
1002                 if (ctx->commits.list[i]->generation != GENERATION_NUMBER_INFINITY &&
1003                     ctx->commits.list[i]->generation != GENERATION_NUMBER_ZERO)
1004                         continue;
1005
1006                 commit_list_insert(ctx->commits.list[i], &list);
1007                 while (list) {
1008                         struct commit *current = list->item;
1009                         struct commit_list *parent;
1010                         int all_parents_computed = 1;
1011                         uint32_t max_generation = 0;
1012
1013                         for (parent = current->parents; parent; parent = parent->next) {
1014                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
1015                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
1016                                         all_parents_computed = 0;
1017                                         commit_list_insert(parent->item, &list);
1018                                         break;
1019                                 } else if (parent->item->generation > max_generation) {
1020                                         max_generation = parent->item->generation;
1021                                 }
1022                         }
1023
1024                         if (all_parents_computed) {
1025                                 current->generation = max_generation + 1;
1026                                 pop_commit(&list);
1027
1028                                 if (current->generation > GENERATION_NUMBER_MAX)
1029                                         current->generation = GENERATION_NUMBER_MAX;
1030                         }
1031                 }
1032         }
1033         stop_progress(&ctx->progress);
1034 }
1035
1036 static int add_ref_to_list(const char *refname,
1037                            const struct object_id *oid,
1038                            int flags, void *cb_data)
1039 {
1040         struct string_list *list = (struct string_list *)cb_data;
1041
1042         string_list_append(list, oid_to_hex(oid));
1043         return 0;
1044 }
1045
1046 int write_commit_graph_reachable(const char *obj_dir, unsigned int flags)
1047 {
1048         struct string_list list = STRING_LIST_INIT_DUP;
1049         int result;
1050
1051         for_each_ref(add_ref_to_list, &list);
1052         result = write_commit_graph(obj_dir, NULL, &list,
1053                                     flags);
1054
1055         string_list_clear(&list, 0);
1056         return result;
1057 }
1058
1059 static int fill_oids_from_packs(struct write_commit_graph_context *ctx,
1060                                 struct string_list *pack_indexes)
1061 {
1062         uint32_t i;
1063         struct strbuf progress_title = STRBUF_INIT;
1064         struct strbuf packname = STRBUF_INIT;
1065         int dirlen;
1066
1067         strbuf_addf(&packname, "%s/pack/", ctx->obj_dir);
1068         dirlen = packname.len;
1069         if (ctx->report_progress) {
1070                 strbuf_addf(&progress_title,
1071                             Q_("Finding commits for commit graph in %d pack",
1072                                "Finding commits for commit graph in %d packs",
1073                                pack_indexes->nr),
1074                             pack_indexes->nr);
1075                 ctx->progress = start_delayed_progress(progress_title.buf, 0);
1076                 ctx->progress_done = 0;
1077         }
1078         for (i = 0; i < pack_indexes->nr; i++) {
1079                 struct packed_git *p;
1080                 strbuf_setlen(&packname, dirlen);
1081                 strbuf_addstr(&packname, pack_indexes->items[i].string);
1082                 p = add_packed_git(packname.buf, packname.len, 1);
1083                 if (!p) {
1084                         error(_("error adding pack %s"), packname.buf);
1085                         return -1;
1086                 }
1087                 if (open_pack_index(p)) {
1088                         error(_("error opening index for %s"), packname.buf);
1089                         return -1;
1090                 }
1091                 for_each_object_in_pack(p, add_packed_commits, ctx,
1092                                         FOR_EACH_OBJECT_PACK_ORDER);
1093                 close_pack(p);
1094                 free(p);
1095         }
1096
1097         stop_progress(&ctx->progress);
1098         strbuf_reset(&progress_title);
1099         strbuf_release(&packname);
1100
1101         return 0;
1102 }
1103
1104 static void fill_oids_from_commit_hex(struct write_commit_graph_context *ctx,
1105                                       struct string_list *commit_hex)
1106 {
1107         uint32_t i;
1108         struct strbuf progress_title = STRBUF_INIT;
1109
1110         if (ctx->report_progress) {
1111                 strbuf_addf(&progress_title,
1112                             Q_("Finding commits for commit graph from %d ref",
1113                                "Finding commits for commit graph from %d refs",
1114                                commit_hex->nr),
1115                             commit_hex->nr);
1116                 ctx->progress = start_delayed_progress(
1117                                         progress_title.buf,
1118                                         commit_hex->nr);
1119         }
1120         for (i = 0; i < commit_hex->nr; i++) {
1121                 const char *end;
1122                 struct object_id oid;
1123                 struct commit *result;
1124
1125                 display_progress(ctx->progress, i + 1);
1126                 if (commit_hex->items[i].string &&
1127                     parse_oid_hex(commit_hex->items[i].string, &oid, &end))
1128                         continue;
1129
1130                 result = lookup_commit_reference_gently(ctx->r, &oid, 1);
1131
1132                 if (result) {
1133                         ALLOC_GROW(ctx->oids.list, ctx->oids.nr + 1, ctx->oids.alloc);
1134                         oidcpy(&ctx->oids.list[ctx->oids.nr], &(result->object.oid));
1135                         ctx->oids.nr++;
1136                 }
1137         }
1138         stop_progress(&ctx->progress);
1139         strbuf_release(&progress_title);
1140 }
1141
1142 static void fill_oids_from_all_packs(struct write_commit_graph_context *ctx)
1143 {
1144         if (ctx->report_progress)
1145                 ctx->progress = start_delayed_progress(
1146                         _("Finding commits for commit graph among packed objects"),
1147                         ctx->approx_nr_objects);
1148         for_each_packed_object(add_packed_commits, ctx,
1149                                FOR_EACH_OBJECT_PACK_ORDER);
1150         if (ctx->progress_done < ctx->approx_nr_objects)
1151                 display_progress(ctx->progress, ctx->approx_nr_objects);
1152         stop_progress(&ctx->progress);
1153 }
1154
1155 static uint32_t count_distinct_commits(struct write_commit_graph_context *ctx)
1156 {
1157         uint32_t i, count_distinct = 1;
1158
1159         if (ctx->report_progress)
1160                 ctx->progress = start_delayed_progress(
1161                         _("Counting distinct commits in commit graph"),
1162                         ctx->oids.nr);
1163         display_progress(ctx->progress, 0); /* TODO: Measure QSORT() progress */
1164         QSORT(ctx->oids.list, ctx->oids.nr, oid_compare);
1165
1166         for (i = 1; i < ctx->oids.nr; i++) {
1167                 display_progress(ctx->progress, i + 1);
1168                 if (!oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1169                         count_distinct++;
1170         }
1171         stop_progress(&ctx->progress);
1172
1173         return count_distinct;
1174 }
1175
1176 static void copy_oids_to_commits(struct write_commit_graph_context *ctx)
1177 {
1178         uint32_t i;
1179         struct commit_list *parent;
1180
1181         ctx->num_extra_edges = 0;
1182         if (ctx->report_progress)
1183                 ctx->progress = start_delayed_progress(
1184                         _("Finding extra edges in commit graph"),
1185                         ctx->oids.nr);
1186         for (i = 0; i < ctx->oids.nr; i++) {
1187                 int num_parents = 0;
1188                 display_progress(ctx->progress, i + 1);
1189                 if (i > 0 && oideq(&ctx->oids.list[i - 1], &ctx->oids.list[i]))
1190                         continue;
1191
1192                 ctx->commits.list[ctx->commits.nr] = lookup_commit(ctx->r, &ctx->oids.list[i]);
1193                 parse_commit_no_graph(ctx->commits.list[ctx->commits.nr]);
1194
1195                 for (parent = ctx->commits.list[ctx->commits.nr]->parents;
1196                      parent; parent = parent->next)
1197                         num_parents++;
1198
1199                 if (num_parents > 2)
1200                         ctx->num_extra_edges += num_parents - 1;
1201
1202                 ctx->commits.nr++;
1203         }
1204         stop_progress(&ctx->progress);
1205 }
1206
1207 static int write_commit_graph_file(struct write_commit_graph_context *ctx)
1208 {
1209         uint32_t i;
1210         struct hashfile *f;
1211         struct lock_file lk = LOCK_INIT;
1212         uint32_t chunk_ids[5];
1213         uint64_t chunk_offsets[5];
1214         const unsigned hashsz = the_hash_algo->rawsz;
1215         struct strbuf progress_title = STRBUF_INIT;
1216         int num_chunks = 3;
1217
1218         ctx->graph_name = get_commit_graph_filename(ctx->obj_dir);
1219         if (safe_create_leading_directories(ctx->graph_name)) {
1220                 UNLEAK(ctx->graph_name);
1221                 error(_("unable to create leading directories of %s"),
1222                         ctx->graph_name);
1223                 return -1;
1224         }
1225
1226         hold_lock_file_for_update(&lk, ctx->graph_name, LOCK_DIE_ON_ERROR);
1227         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1228
1229         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1230         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1231         chunk_ids[2] = GRAPH_CHUNKID_DATA;
1232         if (ctx->num_extra_edges) {
1233                 chunk_ids[num_chunks] = GRAPH_CHUNKID_EXTRAEDGES;
1234                 num_chunks++;
1235         }
1236
1237         chunk_ids[num_chunks] = 0;
1238
1239         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1240         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1241         chunk_offsets[2] = chunk_offsets[1] + hashsz * ctx->commits.nr;
1242         chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * ctx->commits.nr;
1243
1244         num_chunks = 3;
1245         if (ctx->num_extra_edges) {
1246                 chunk_offsets[num_chunks + 1] = chunk_offsets[num_chunks] +
1247                                                 4 * ctx->num_extra_edges;
1248                 num_chunks++;
1249         }
1250
1251         hashwrite_be32(f, GRAPH_SIGNATURE);
1252
1253         hashwrite_u8(f, GRAPH_VERSION);
1254         hashwrite_u8(f, oid_version());
1255         hashwrite_u8(f, num_chunks);
1256         hashwrite_u8(f, 0);
1257
1258         for (i = 0; i <= num_chunks; i++) {
1259                 uint32_t chunk_write[3];
1260
1261                 chunk_write[0] = htonl(chunk_ids[i]);
1262                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1263                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1264                 hashwrite(f, chunk_write, 12);
1265         }
1266
1267         if (ctx->report_progress) {
1268                 strbuf_addf(&progress_title,
1269                             Q_("Writing out commit graph in %d pass",
1270                                "Writing out commit graph in %d passes",
1271                                num_chunks),
1272                             num_chunks);
1273                 ctx->progress = start_delayed_progress(
1274                         progress_title.buf,
1275                         num_chunks * ctx->commits.nr);
1276         }
1277         write_graph_chunk_fanout(f, ctx);
1278         write_graph_chunk_oids(f, hashsz, ctx);
1279         write_graph_chunk_data(f, hashsz, ctx);
1280         if (ctx->num_extra_edges)
1281                 write_graph_chunk_extra_edges(f, ctx);
1282         stop_progress(&ctx->progress);
1283         strbuf_release(&progress_title);
1284
1285         close_commit_graph(ctx->r->objects);
1286         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1287         commit_lock_file(&lk);
1288
1289         return 0;
1290 }
1291
1292 int write_commit_graph(const char *obj_dir,
1293                        struct string_list *pack_indexes,
1294                        struct string_list *commit_hex,
1295                        unsigned int flags)
1296 {
1297         struct write_commit_graph_context *ctx;
1298         uint32_t i, count_distinct = 0;
1299         int res = 0;
1300
1301         if (!commit_graph_compatible(the_repository))
1302                 return 0;
1303
1304         ctx = xcalloc(1, sizeof(struct write_commit_graph_context));
1305         ctx->r = the_repository;
1306         ctx->obj_dir = obj_dir;
1307         ctx->append = flags & COMMIT_GRAPH_APPEND ? 1 : 0;
1308         ctx->report_progress = flags & COMMIT_GRAPH_PROGRESS ? 1 : 0;
1309
1310         ctx->approx_nr_objects = approximate_object_count();
1311         ctx->oids.alloc = ctx->approx_nr_objects / 32;
1312
1313         if (ctx->append) {
1314                 prepare_commit_graph_one(ctx->r, ctx->obj_dir);
1315                 if (ctx->r->objects->commit_graph)
1316                         ctx->oids.alloc += ctx->r->objects->commit_graph->num_commits;
1317         }
1318
1319         if (ctx->oids.alloc < 1024)
1320                 ctx->oids.alloc = 1024;
1321         ALLOC_ARRAY(ctx->oids.list, ctx->oids.alloc);
1322
1323         if (ctx->append && ctx->r->objects->commit_graph) {
1324                 struct commit_graph *g = ctx->r->objects->commit_graph;
1325                 for (i = 0; i < g->num_commits; i++) {
1326                         const unsigned char *hash = g->chunk_oid_lookup + g->hash_len * i;
1327                         hashcpy(ctx->oids.list[ctx->oids.nr++].hash, hash);
1328                 }
1329         }
1330
1331         if (pack_indexes) {
1332                 if ((res = fill_oids_from_packs(ctx, pack_indexes)))
1333                         goto cleanup;
1334         }
1335
1336         if (commit_hex)
1337                 fill_oids_from_commit_hex(ctx, commit_hex);
1338
1339         if (!pack_indexes && !commit_hex)
1340                 fill_oids_from_all_packs(ctx);
1341
1342         close_reachable(ctx);
1343
1344         count_distinct = count_distinct_commits(ctx);
1345
1346         if (count_distinct >= GRAPH_EDGE_LAST_MASK) {
1347                 error(_("the commit graph format cannot write %d commits"), count_distinct);
1348                 res = -1;
1349                 goto cleanup;
1350         }
1351
1352         ctx->commits.alloc = count_distinct;
1353         ALLOC_ARRAY(ctx->commits.list, ctx->commits.alloc);
1354
1355         copy_oids_to_commits(ctx);
1356
1357         if (ctx->commits.nr >= GRAPH_EDGE_LAST_MASK) {
1358                 error(_("too many commits to write graph"));
1359                 res = -1;
1360                 goto cleanup;
1361         }
1362
1363         compute_generation_numbers(ctx);
1364
1365         res = write_commit_graph_file(ctx);
1366
1367 cleanup:
1368         free(ctx->graph_name);
1369         free(ctx->commits.list);
1370         free(ctx->oids.list);
1371         free(ctx);
1372
1373         return res;
1374 }
1375
1376 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1377 static int verify_commit_graph_error;
1378
1379 static void graph_report(const char *fmt, ...)
1380 {
1381         va_list ap;
1382
1383         verify_commit_graph_error = 1;
1384         va_start(ap, fmt);
1385         vfprintf(stderr, fmt, ap);
1386         fprintf(stderr, "\n");
1387         va_end(ap);
1388 }
1389
1390 #define GENERATION_ZERO_EXISTS 1
1391 #define GENERATION_NUMBER_EXISTS 2
1392
1393 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1394 {
1395         uint32_t i, cur_fanout_pos = 0;
1396         struct object_id prev_oid, cur_oid, checksum;
1397         int generation_zero = 0;
1398         struct hashfile *f;
1399         int devnull;
1400         struct progress *progress = NULL;
1401
1402         if (!g) {
1403                 graph_report("no commit-graph file loaded");
1404                 return 1;
1405         }
1406
1407         verify_commit_graph_error = verify_commit_graph_lite(g);
1408         if (verify_commit_graph_error)
1409                 return verify_commit_graph_error;
1410
1411         devnull = open("/dev/null", O_WRONLY);
1412         f = hashfd(devnull, NULL);
1413         hashwrite(f, g->data, g->data_len - g->hash_len);
1414         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1415         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1416                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1417                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1418         }
1419
1420         for (i = 0; i < g->num_commits; i++) {
1421                 struct commit *graph_commit;
1422
1423                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1424
1425                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1426                         graph_report(_("commit-graph has incorrect OID order: %s then %s"),
1427                                      oid_to_hex(&prev_oid),
1428                                      oid_to_hex(&cur_oid));
1429
1430                 oidcpy(&prev_oid, &cur_oid);
1431
1432                 while (cur_oid.hash[0] > cur_fanout_pos) {
1433                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1434
1435                         if (i != fanout_value)
1436                                 graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1437                                              cur_fanout_pos, fanout_value, i);
1438                         cur_fanout_pos++;
1439                 }
1440
1441                 graph_commit = lookup_commit(r, &cur_oid);
1442                 if (!parse_commit_in_graph_one(r, g, graph_commit))
1443                         graph_report(_("failed to parse commit %s from commit-graph"),
1444                                      oid_to_hex(&cur_oid));
1445         }
1446
1447         while (cur_fanout_pos < 256) {
1448                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1449
1450                 if (g->num_commits != fanout_value)
1451                         graph_report(_("commit-graph has incorrect fanout value: fanout[%d] = %u != %u"),
1452                                      cur_fanout_pos, fanout_value, i);
1453
1454                 cur_fanout_pos++;
1455         }
1456
1457         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1458                 return verify_commit_graph_error;
1459
1460         progress = start_progress(_("Verifying commits in commit graph"),
1461                                   g->num_commits);
1462         for (i = 0; i < g->num_commits; i++) {
1463                 struct commit *graph_commit, *odb_commit;
1464                 struct commit_list *graph_parents, *odb_parents;
1465                 uint32_t max_generation = 0;
1466
1467                 display_progress(progress, i + 1);
1468                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1469
1470                 graph_commit = lookup_commit(r, &cur_oid);
1471                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1472                 if (parse_commit_internal(odb_commit, 0, 0)) {
1473                         graph_report(_("failed to parse commit %s from object database for commit-graph"),
1474                                      oid_to_hex(&cur_oid));
1475                         continue;
1476                 }
1477
1478                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1479                            get_commit_tree_oid(odb_commit)))
1480                         graph_report(_("root tree OID for commit %s in commit-graph is %s != %s"),
1481                                      oid_to_hex(&cur_oid),
1482                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1483                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1484
1485                 graph_parents = graph_commit->parents;
1486                 odb_parents = odb_commit->parents;
1487
1488                 while (graph_parents) {
1489                         if (odb_parents == NULL) {
1490                                 graph_report(_("commit-graph parent list for commit %s is too long"),
1491                                              oid_to_hex(&cur_oid));
1492                                 break;
1493                         }
1494
1495                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1496                                 graph_report(_("commit-graph parent for %s is %s != %s"),
1497                                              oid_to_hex(&cur_oid),
1498                                              oid_to_hex(&graph_parents->item->object.oid),
1499                                              oid_to_hex(&odb_parents->item->object.oid));
1500
1501                         if (graph_parents->item->generation > max_generation)
1502                                 max_generation = graph_parents->item->generation;
1503
1504                         graph_parents = graph_parents->next;
1505                         odb_parents = odb_parents->next;
1506                 }
1507
1508                 if (odb_parents != NULL)
1509                         graph_report(_("commit-graph parent list for commit %s terminates early"),
1510                                      oid_to_hex(&cur_oid));
1511
1512                 if (!graph_commit->generation) {
1513                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1514                                 graph_report(_("commit-graph has generation number zero for commit %s, but non-zero elsewhere"),
1515                                              oid_to_hex(&cur_oid));
1516                         generation_zero = GENERATION_ZERO_EXISTS;
1517                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1518                         graph_report(_("commit-graph has non-zero generation number for commit %s, but zero elsewhere"),
1519                                      oid_to_hex(&cur_oid));
1520
1521                 if (generation_zero == GENERATION_ZERO_EXISTS)
1522                         continue;
1523
1524                 /*
1525                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1526                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1527                  * extra logic in the following condition.
1528                  */
1529                 if (max_generation == GENERATION_NUMBER_MAX)
1530                         max_generation--;
1531
1532                 if (graph_commit->generation != max_generation + 1)
1533                         graph_report(_("commit-graph generation for commit %s is %u != %u"),
1534                                      oid_to_hex(&cur_oid),
1535                                      graph_commit->generation,
1536                                      max_generation + 1);
1537
1538                 if (graph_commit->date != odb_commit->date)
1539                         graph_report(_("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime),
1540                                      oid_to_hex(&cur_oid),
1541                                      graph_commit->date,
1542                                      odb_commit->date);
1543         }
1544         stop_progress(&progress);
1545
1546         return verify_commit_graph_error;
1547 }
1548
1549 void free_commit_graph(struct commit_graph *g)
1550 {
1551         if (!g)
1552                 return;
1553         if (g->graph_fd >= 0) {
1554                 munmap((void *)g->data, g->data_len);
1555                 g->data = NULL;
1556                 close(g->graph_fd);
1557         }
1558         free(g);
1559 }