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
  17 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  18 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  19 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  20 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  21 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  22
  23 #define GRAPH_DATA_WIDTH 36
  24
  25 #define GRAPH_VERSION_1 0x1
  26 #define GRAPH_VERSION GRAPH_VERSION_1
  27
  28 #define GRAPH_OID_VERSION_SHA1 1
  29 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
  30 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
  31 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
  32
  33 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
  34 #define GRAPH_PARENT_MISSING 0x7fffffff
  35 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  36 #define GRAPH_PARENT_NONE 0x70000000
  37
  38 #define GRAPH_LAST_EDGE 0x80000000
  39
  40 #define GRAPH_HEADER_SIZE 8
  41 #define GRAPH_FANOUT_SIZE (4 * 256)
  42 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  43 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  44                         + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
  45
  46 char *get_commit_graph_filename(const char *obj_dir)
  47 {
  48         return xstrfmt("%s/info/commit-graph", obj_dir);
  49 }
  50
  51 static struct commit_graph *alloc_commit_graph(void)
  52 {
  53         struct commit_graph *g = xcalloc(1, sizeof(*g));
  54         g->graph_fd = -1;
  55
  56         return g;
  57 }
  58
  59 struct commit_graph *load_commit_graph_one(const char *graph_file)
  60 {
  61         void *graph_map;
  62         const unsigned char *data, *chunk_lookup;
  63         size_t graph_size;
  64         struct stat st;
  65         uint32_t i;
  66         struct commit_graph *graph;
  67         int fd = git_open(graph_file);
  68         uint64_t last_chunk_offset;
  69         uint32_t last_chunk_id;
  70         uint32_t graph_signature;
  71         unsigned char graph_version, hash_version;
  72
  73         if (fd < 0)
  74                 return NULL;
  75         if (fstat(fd, &st)) {
  76                 close(fd);
  77                 return NULL;
  78         }
  79         graph_size = xsize_t(st.st_size);
  80
  81         if (graph_size < GRAPH_MIN_SIZE) {
  82                 close(fd);
  83                 die("graph file %s is too small", graph_file);
  84         }
  85         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
  86         data = (const unsigned char *)graph_map;
  87
  88         graph_signature = get_be32(data);
  89         if (graph_signature != GRAPH_SIGNATURE) {
  90                 error("graph signature %X does not match signature %X",
  91                       graph_signature, GRAPH_SIGNATURE);
  92                 goto cleanup_fail;
  93         }
  94
  95         graph_version = *(unsigned char*)(data + 4);
  96         if (graph_version != GRAPH_VERSION) {
  97                 error("graph version %X does not match version %X",
  98                       graph_version, GRAPH_VERSION);
  99                 goto cleanup_fail;
 100         }
 101
 102         hash_version = *(unsigned char*)(data + 5);
 103         if (hash_version != GRAPH_OID_VERSION) {
 104                 error("hash version %X does not match version %X",
 105                       hash_version, GRAPH_OID_VERSION);
 106                 goto cleanup_fail;
 107         }
 108
 109         graph = alloc_commit_graph();
 110
 111         graph->hash_len = GRAPH_OID_LEN;
 112         graph->num_chunks = *(unsigned char*)(data + 6);
 113         graph->graph_fd = fd;
 114         graph->data = graph_map;
 115         graph->data_len = graph_size;
 116
 117         last_chunk_id = 0;
 118         last_chunk_offset = 8;
 119         chunk_lookup = data + 8;
 120         for (i = 0; i < graph->num_chunks; i++) {
 121                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 122                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 123                 int chunk_repeated = 0;
 124
 125                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 126
 127                 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
 128                         error("improper chunk offset %08x%08x", (uint32_t)(chunk_offset >> 32),
 129                               (uint32_t)chunk_offset);
 130                         goto cleanup_fail;
 131                 }
 132
 133                 switch (chunk_id) {
 134                 case GRAPH_CHUNKID_OIDFANOUT:
 135                         if (graph->chunk_oid_fanout)
 136                                 chunk_repeated = 1;
 137                         else
 138                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 139                         break;
 140
 141                 case GRAPH_CHUNKID_OIDLOOKUP:
 142                         if (graph->chunk_oid_lookup)
 143                                 chunk_repeated = 1;
 144                         else
 145                                 graph->chunk_oid_lookup = data + chunk_offset;
 146                         break;
 147
 148                 case GRAPH_CHUNKID_DATA:
 149                         if (graph->chunk_commit_data)
 150                                 chunk_repeated = 1;
 151                         else
 152                                 graph->chunk_commit_data = data + chunk_offset;
 153                         break;
 154
 155                 case GRAPH_CHUNKID_LARGEEDGES:
 156                         if (graph->chunk_large_edges)
 157                                 chunk_repeated = 1;
 158                         else
 159                                 graph->chunk_large_edges = data + chunk_offset;
 160                         break;
 161                 }
 162
 163                 if (chunk_repeated) {
 164                         error("chunk id %08x appears multiple times", chunk_id);
 165                         goto cleanup_fail;
 166                 }
 167
 168                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 169                 {
 170                         graph->num_commits = (chunk_offset - last_chunk_offset)
 171                                              / graph->hash_len;
 172                 }
 173
 174                 last_chunk_id = chunk_id;
 175                 last_chunk_offset = chunk_offset;
 176         }
 177
 178         return graph;
 179
 180 cleanup_fail:
 181         munmap(graph_map, graph_size);
 182         close(fd);
 183         exit(1);
 184 }
 185
 186 /* global storage */
 187 static struct commit_graph *commit_graph = NULL;
 188
 189 static void prepare_commit_graph_one(const char *obj_dir)
 190 {
 191         char *graph_name;
 192
 193         if (commit_graph)
 194                 return;
 195
 196         graph_name = get_commit_graph_filename(obj_dir);
 197         commit_graph = load_commit_graph_one(graph_name);
 198
 199         FREE_AND_NULL(graph_name);
 200 }
 201
 202 static int prepare_commit_graph_run_once = 0;
 203
 204 /*
 205  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 206  *
 207  * On the first invocation, this function attemps to load the commit
 208  * graph if the_repository is configured to have one.
 209  */
 210 static int prepare_commit_graph(void)
 211 {
 212         struct alternate_object_database *alt;
 213         char *obj_dir;
 214
 215         if (prepare_commit_graph_run_once)
 216                 return !!commit_graph;
 217         prepare_commit_graph_run_once = 1;
 218
 219         if (!core_commit_graph)
 220                 return 0;
 221
 222         obj_dir = get_object_directory();
 223         prepare_commit_graph_one(obj_dir);
 224         prepare_alt_odb(the_repository);
 225         for (alt = the_repository->objects->alt_odb_list;
 226              !commit_graph && alt;
 227              alt = alt->next)
 228                 prepare_commit_graph_one(alt->path);
 229         return !!commit_graph;
 230 }
 231
 232 static void close_commit_graph(void)
 233 {
 234         free_commit_graph(commit_graph);
 235         commit_graph = NULL;
 236 }
 237
 238 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 239 {
 240         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 241                             g->chunk_oid_lookup, g->hash_len, pos);
 242 }
 243
 244 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 245                                                  uint64_t pos,
 246                                                  struct commit_list **pptr)
 247 {
 248         struct commit *c;
 249         struct object_id oid;
 250
 251         if (pos >= g->num_commits)
 252                 die("invalid parent position %"PRIu64, pos);
 253
 254         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 255         c = lookup_commit(the_repository, &oid);
 256         if (!c)
 257                 die("could not find commit %s", oid_to_hex(&oid));
 258         c->graph_pos = pos;
 259         return &commit_list_insert(c, pptr)->next;
 260 }
 261
 262 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 263 {
 264         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 265         item->graph_pos = pos;
 266         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 267 }
 268
 269 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 270 {
 271         uint32_t edge_value;
 272         uint32_t *parent_data_ptr;
 273         uint64_t date_low, date_high;
 274         struct commit_list **pptr;
 275         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 276
 277         item->object.parsed = 1;
 278         item->graph_pos = pos;
 279
 280         item->maybe_tree = NULL;
 281
 282         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 283         date_low = get_be32(commit_data + g->hash_len + 12);
 284         item->date = (timestamp_t)((date_high << 32) | date_low);
 285
 286         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 287
 288         pptr = &item->parents;
 289
 290         edge_value = get_be32(commit_data + g->hash_len);
 291         if (edge_value == GRAPH_PARENT_NONE)
 292                 return 1;
 293         pptr = insert_parent_or_die(g, edge_value, pptr);
 294
 295         edge_value = get_be32(commit_data + g->hash_len + 4);
 296         if (edge_value == GRAPH_PARENT_NONE)
 297                 return 1;
 298         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 299                 pptr = insert_parent_or_die(g, edge_value, pptr);
 300                 return 1;
 301         }
 302
 303         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 304                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 305         do {
 306                 edge_value = get_be32(parent_data_ptr);
 307                 pptr = insert_parent_or_die(g,
 308                                             edge_value & GRAPH_EDGE_LAST_MASK,
 309                                             pptr);
 310                 parent_data_ptr++;
 311         } while (!(edge_value & GRAPH_LAST_EDGE));
 312
 313         return 1;
 314 }
 315
 316 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 317 {
 318         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 319                 *pos = item->graph_pos;
 320                 return 1;
 321         } else {
 322                 return bsearch_graph(g, &(item->object.oid), pos);
 323         }
 324 }
 325
 326 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 327 {
 328         uint32_t pos;
 329
 330         if (!core_commit_graph)
 331                 return 0;
 332         if (item->object.parsed)
 333                 return 1;
 334
 335         if (find_commit_in_graph(item, g, &pos))
 336                 return fill_commit_in_graph(item, g, pos);
 337
 338         return 0;
 339 }
 340
 341 int parse_commit_in_graph(struct commit *item)
 342 {
 343         if (!prepare_commit_graph())
 344                 return 0;
 345         return parse_commit_in_graph_one(commit_graph, item);
 346 }
 347
 348 void load_commit_graph_info(struct commit *item)
 349 {
 350         uint32_t pos;
 351         if (!prepare_commit_graph())
 352                 return;
 353         if (find_commit_in_graph(item, commit_graph, &pos))
 354                 fill_commit_graph_info(item, commit_graph, pos);
 355 }
 356
 357 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 358 {
 359         struct object_id oid;
 360         const unsigned char *commit_data = g->chunk_commit_data +
 361                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 362
 363         hashcpy(oid.hash, commit_data);
 364         c->maybe_tree = lookup_tree(the_repository, &oid);
 365
 366         return c->maybe_tree;
 367 }
 368
 369 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 370                                                  const struct commit *c)
 371 {
 372         if (c->maybe_tree)
 373                 return c->maybe_tree;
 374         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 375                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 376
 377         return load_tree_for_commit(g, (struct commit *)c);
 378 }
 379
 380 struct tree *get_commit_tree_in_graph(const struct commit *c)
 381 {
 382         return get_commit_tree_in_graph_one(commit_graph, c);
 383 }
 384
 385 static void write_graph_chunk_fanout(struct hashfile *f,
 386                                      struct commit **commits,
 387                                      int nr_commits)
 388 {
 389         int i, count = 0;
 390         struct commit **list = commits;
 391
 392         /*
 393          * Write the first-level table (the list is sorted,
 394          * but we use a 256-entry lookup to be able to avoid
 395          * having to do eight extra binary search iterations).
 396          */
 397         for (i = 0; i < 256; i++) {
 398                 while (count < nr_commits) {
 399                         if ((*list)->object.oid.hash[0] != i)
 400                                 break;
 401                         count++;
 402                         list++;
 403                 }
 404
 405                 hashwrite_be32(f, count);
 406         }
 407 }
 408
 409 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 410                                    struct commit **commits, int nr_commits)
 411 {
 412         struct commit **list = commits;
 413         int count;
 414         for (count = 0; count < nr_commits; count++, list++)
 415                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 416 }
 417
 418 static const unsigned char *commit_to_sha1(size_t index, void *table)
 419 {
 420         struct commit **commits = table;
 421         return commits[index]->object.oid.hash;
 422 }
 423
 424 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 425                                    struct commit **commits, int nr_commits)
 426 {
 427         struct commit **list = commits;
 428         struct commit **last = commits + nr_commits;
 429         uint32_t num_extra_edges = 0;
 430
 431         while (list < last) {
 432                 struct commit_list *parent;
 433                 int edge_value;
 434                 uint32_t packedDate[2];
 435
 436                 parse_commit(*list);
 437                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 438
 439                 parent = (*list)->parents;
 440
 441                 if (!parent)
 442                         edge_value = GRAPH_PARENT_NONE;
 443                 else {
 444                         edge_value = sha1_pos(parent->item->object.oid.hash,
 445                                               commits,
 446                                               nr_commits,
 447                                               commit_to_sha1);
 448
 449                         if (edge_value < 0)
 450                                 edge_value = GRAPH_PARENT_MISSING;
 451                 }
 452
 453                 hashwrite_be32(f, edge_value);
 454
 455                 if (parent)
 456                         parent = parent->next;
 457
 458                 if (!parent)
 459                         edge_value = GRAPH_PARENT_NONE;
 460                 else if (parent->next)
 461                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 462                 else {
 463                         edge_value = sha1_pos(parent->item->object.oid.hash,
 464                                               commits,
 465                                               nr_commits,
 466                                               commit_to_sha1);
 467                         if (edge_value < 0)
 468                                 edge_value = GRAPH_PARENT_MISSING;
 469                 }
 470
 471                 hashwrite_be32(f, edge_value);
 472
 473                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 474                         do {
 475                                 num_extra_edges++;
 476                                 parent = parent->next;
 477                         } while (parent);
 478                 }
 479
 480                 if (sizeof((*list)->date) > 4)
 481                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 482                 else
 483                         packedDate[0] = 0;
 484
 485                 packedDate[0] |= htonl((*list)->generation << 2);
 486
 487                 packedDate[1] = htonl((*list)->date);
 488                 hashwrite(f, packedDate, 8);
 489
 490                 list++;
 491         }
 492 }
 493
 494 static void write_graph_chunk_large_edges(struct hashfile *f,
 495                                           struct commit **commits,
 496                                           int nr_commits)
 497 {
 498         struct commit **list = commits;
 499         struct commit **last = commits + nr_commits;
 500         struct commit_list *parent;
 501
 502         while (list < last) {
 503                 int num_parents = 0;
 504                 for (parent = (*list)->parents; num_parents < 3 && parent;
 505                      parent = parent->next)
 506                         num_parents++;
 507
 508                 if (num_parents <= 2) {
 509                         list++;
 510                         continue;
 511                 }
 512
 513                 /* Since num_parents > 2, this initializer is safe. */
 514                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 515                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 516                                                   commits,
 517                                                   nr_commits,
 518                                                   commit_to_sha1);
 519
 520                         if (edge_value < 0)
 521                                 edge_value = GRAPH_PARENT_MISSING;
 522                         else if (!parent->next)
 523                                 edge_value |= GRAPH_LAST_EDGE;
 524
 525                         hashwrite_be32(f, edge_value);
 526                 }
 527
 528                 list++;
 529         }
 530 }
 531
 532 static int commit_compare(const void *_a, const void *_b)
 533 {
 534         const struct object_id *a = (const struct object_id *)_a;
 535         const struct object_id *b = (const struct object_id *)_b;
 536         return oidcmp(a, b);
 537 }
 538
 539 struct packed_commit_list {
 540         struct commit **list;
 541         int nr;
 542         int alloc;
 543 };
 544
 545 struct packed_oid_list {
 546         struct object_id *list;
 547         int nr;
 548         int alloc;
 549 };
 550
 551 static int add_packed_commits(const struct object_id *oid,
 552                               struct packed_git *pack,
 553                               uint32_t pos,
 554                               void *data)
 555 {
 556         struct packed_oid_list *list = (struct packed_oid_list*)data;
 557         enum object_type type;
 558         off_t offset = nth_packed_object_offset(pack, pos);
 559         struct object_info oi = OBJECT_INFO_INIT;
 560
 561         oi.typep = &type;
 562         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 563                 die("unable to get type of object %s", oid_to_hex(oid));
 564
 565         if (type != OBJ_COMMIT)
 566                 return 0;
 567
 568         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 569         oidcpy(&(list->list[list->nr]), oid);
 570         list->nr++;
 571
 572         return 0;
 573 }
 574
 575 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 576 {
 577         struct commit_list *parent;
 578         for (parent = commit->parents; parent; parent = parent->next) {
 579                 if (!(parent->item->object.flags & UNINTERESTING)) {
 580                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 581                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 582                         oids->nr++;
 583                         parent->item->object.flags |= UNINTERESTING;
 584                 }
 585         }
 586 }
 587
 588 static void close_reachable(struct packed_oid_list *oids)
 589 {
 590         int i;
 591         struct commit *commit;
 592
 593         for (i = 0; i < oids->nr; i++) {
 594                 commit = lookup_commit(the_repository, &oids->list[i]);
 595                 if (commit)
 596                         commit->object.flags |= UNINTERESTING;
 597         }
 598
 599         /*
 600          * As this loop runs, oids->nr may grow, but not more
 601          * than the number of missing commits in the reachable
 602          * closure.
 603          */
 604         for (i = 0; i < oids->nr; i++) {
 605                 commit = lookup_commit(the_repository, &oids->list[i]);
 606
 607                 if (commit && !parse_commit(commit))
 608                         add_missing_parents(oids, commit);
 609         }
 610
 611         for (i = 0; i < oids->nr; i++) {
 612                 commit = lookup_commit(the_repository, &oids->list[i]);
 613
 614                 if (commit)
 615                         commit->object.flags &= ~UNINTERESTING;
 616         }
 617 }
 618
 619 static void compute_generation_numbers(struct packed_commit_list* commits)
 620 {
 621         int i;
 622         struct commit_list *list = NULL;
 623
 624         for (i = 0; i < commits->nr; i++) {
 625                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 626                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 627                         continue;
 628
 629                 commit_list_insert(commits->list[i], &list);
 630                 while (list) {
 631                         struct commit *current = list->item;
 632                         struct commit_list *parent;
 633                         int all_parents_computed = 1;
 634                         uint32_t max_generation = 0;
 635
 636                         for (parent = current->parents; parent; parent = parent->next) {
 637                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 638                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 639                                         all_parents_computed = 0;
 640                                         commit_list_insert(parent->item, &list);
 641                                         break;
 642                                 } else if (parent->item->generation > max_generation) {
 643                                         max_generation = parent->item->generation;
 644                                 }
 645                         }
 646
 647                         if (all_parents_computed) {
 648                                 current->generation = max_generation + 1;
 649                                 pop_commit(&list);
 650
 651                                 if (current->generation > GENERATION_NUMBER_MAX)
 652                                         current->generation = GENERATION_NUMBER_MAX;
 653                         }
 654                 }
 655         }
 656 }
 657
 658 static int add_ref_to_list(const char *refname,
 659                            const struct object_id *oid,
 660                            int flags, void *cb_data)
 661 {
 662         struct string_list *list = (struct string_list *)cb_data;
 663
 664         string_list_append(list, oid_to_hex(oid));
 665         return 0;
 666 }
 667
 668 void write_commit_graph_reachable(const char *obj_dir, int append)
 669 {
 670         struct string_list list;
 671
 672         string_list_init(&list, 1);
 673         for_each_ref(add_ref_to_list, &list);
 674         write_commit_graph(obj_dir, NULL, &list, append);
 675 }
 676
 677 void write_commit_graph(const char *obj_dir,
 678                         struct string_list *pack_indexes,
 679                         struct string_list *commit_hex,
 680                         int append)
 681 {
 682         struct packed_oid_list oids;
 683         struct packed_commit_list commits;
 684         struct hashfile *f;
 685         uint32_t i, count_distinct = 0;
 686         char *graph_name;
 687         struct lock_file lk = LOCK_INIT;
 688         uint32_t chunk_ids[5];
 689         uint64_t chunk_offsets[5];
 690         int num_chunks;
 691         int num_extra_edges;
 692         struct commit_list *parent;
 693
 694         oids.nr = 0;
 695         oids.alloc = approximate_object_count() / 4;
 696
 697         if (append) {
 698                 prepare_commit_graph_one(obj_dir);
 699                 if (commit_graph)
 700                         oids.alloc += commit_graph->num_commits;
 701         }
 702
 703         if (oids.alloc < 1024)
 704                 oids.alloc = 1024;
 705         ALLOC_ARRAY(oids.list, oids.alloc);
 706
 707         if (append && commit_graph) {
 708                 for (i = 0; i < commit_graph->num_commits; i++) {
 709                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 710                                 commit_graph->hash_len * i;
 711                         hashcpy(oids.list[oids.nr++].hash, hash);
 712                 }
 713         }
 714
 715         if (pack_indexes) {
 716                 struct strbuf packname = STRBUF_INIT;
 717                 int dirlen;
 718                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 719                 dirlen = packname.len;
 720                 for (i = 0; i < pack_indexes->nr; i++) {
 721                         struct packed_git *p;
 722                         strbuf_setlen(&packname, dirlen);
 723                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 724                         p = add_packed_git(packname.buf, packname.len, 1);
 725                         if (!p)
 726                                 die("error adding pack %s", packname.buf);
 727                         if (open_pack_index(p))
 728                                 die("error opening index for %s", packname.buf);
 729                         for_each_object_in_pack(p, add_packed_commits, &oids);
 730                         close_pack(p);
 731                 }
 732                 strbuf_release(&packname);
 733         }
 734
 735         if (commit_hex) {
 736                 for (i = 0; i < commit_hex->nr; i++) {
 737                         const char *end;
 738                         struct object_id oid;
 739                         struct commit *result;
 740
 741                         if (commit_hex->items[i].string &&
 742                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 743                                 continue;
 744
 745                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
 746
 747                         if (result) {
 748                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 749                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 750                                 oids.nr++;
 751                         }
 752                 }
 753         }
 754
 755         if (!pack_indexes && !commit_hex)
 756                 for_each_packed_object(add_packed_commits, &oids, 0);
 757
 758         close_reachable(&oids);
 759
 760         QSORT(oids.list, oids.nr, commit_compare);
 761
 762         count_distinct = 1;
 763         for (i = 1; i < oids.nr; i++) {
 764                 if (oidcmp(&oids.list[i-1], &oids.list[i]))
 765                         count_distinct++;
 766         }
 767
 768         if (count_distinct >= GRAPH_PARENT_MISSING)
 769                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 770
 771         commits.nr = 0;
 772         commits.alloc = count_distinct;
 773         ALLOC_ARRAY(commits.list, commits.alloc);
 774
 775         num_extra_edges = 0;
 776         for (i = 0; i < oids.nr; i++) {
 777                 int num_parents = 0;
 778                 if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
 779                         continue;
 780
 781                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
 782                 parse_commit(commits.list[commits.nr]);
 783
 784                 for (parent = commits.list[commits.nr]->parents;
 785                      parent; parent = parent->next)
 786                         num_parents++;
 787
 788                 if (num_parents > 2)
 789                         num_extra_edges += num_parents - 1;
 790
 791                 commits.nr++;
 792         }
 793         num_chunks = num_extra_edges ? 4 : 3;
 794
 795         if (commits.nr >= GRAPH_PARENT_MISSING)
 796                 die(_("too many commits to write graph"));
 797
 798         compute_generation_numbers(&commits);
 799
 800         graph_name = get_commit_graph_filename(obj_dir);
 801         if (safe_create_leading_directories(graph_name))
 802                 die_errno(_("unable to create leading directories of %s"),
 803                           graph_name);
 804
 805         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 806         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 807
 808         hashwrite_be32(f, GRAPH_SIGNATURE);
 809
 810         hashwrite_u8(f, GRAPH_VERSION);
 811         hashwrite_u8(f, GRAPH_OID_VERSION);
 812         hashwrite_u8(f, num_chunks);
 813         hashwrite_u8(f, 0); /* unused padding byte */
 814
 815         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 816         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 817         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 818         if (num_extra_edges)
 819                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 820         else
 821                 chunk_ids[3] = 0;
 822         chunk_ids[4] = 0;
 823
 824         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 825         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 826         chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 827         chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 828         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 829
 830         for (i = 0; i <= num_chunks; i++) {
 831                 uint32_t chunk_write[3];
 832
 833                 chunk_write[0] = htonl(chunk_ids[i]);
 834                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 835                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 836                 hashwrite(f, chunk_write, 12);
 837         }
 838
 839         write_graph_chunk_fanout(f, commits.list, commits.nr);
 840         write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 841         write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 842         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 843
 844         close_commit_graph();
 845         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 846         commit_lock_file(&lk);
 847
 848         free(oids.list);
 849         oids.alloc = 0;
 850         oids.nr = 0;
 851 }
 852
 853 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 854 static int verify_commit_graph_error;
 855
 856 static void graph_report(const char *fmt, ...)
 857 {
 858         va_list ap;
 859
 860         verify_commit_graph_error = 1;
 861         va_start(ap, fmt);
 862         vfprintf(stderr, fmt, ap);
 863         fprintf(stderr, "\n");
 864         va_end(ap);
 865 }
 866
 867 #define GENERATION_ZERO_EXISTS 1
 868 #define GENERATION_NUMBER_EXISTS 2
 869
 870 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 871 {
 872         uint32_t i, cur_fanout_pos = 0;
 873         struct object_id prev_oid, cur_oid, checksum;
 874         int generation_zero = 0;
 875         struct hashfile *f;
 876         int devnull;
 877
 878         if (!g) {
 879                 graph_report("no commit-graph file loaded");
 880                 return 1;
 881         }
 882
 883         verify_commit_graph_error = 0;
 884
 885         if (!g->chunk_oid_fanout)
 886                 graph_report("commit-graph is missing the OID Fanout chunk");
 887         if (!g->chunk_oid_lookup)
 888                 graph_report("commit-graph is missing the OID Lookup chunk");
 889         if (!g->chunk_commit_data)
 890                 graph_report("commit-graph is missing the Commit Data chunk");
 891
 892         if (verify_commit_graph_error)
 893                 return verify_commit_graph_error;
 894
 895         devnull = open("/dev/null", O_WRONLY);
 896         f = hashfd(devnull, NULL);
 897         hashwrite(f, g->data, g->data_len - g->hash_len);
 898         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
 899         if (hashcmp(checksum.hash, g->data + g->data_len - g->hash_len)) {
 900                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
 901                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
 902         }
 903
 904         for (i = 0; i < g->num_commits; i++) {
 905                 struct commit *graph_commit;
 906
 907                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 908
 909                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
 910                         graph_report("commit-graph has incorrect OID order: %s then %s",
 911                                      oid_to_hex(&prev_oid),
 912                                      oid_to_hex(&cur_oid));
 913
 914                 oidcpy(&prev_oid, &cur_oid);
 915
 916                 while (cur_oid.hash[0] > cur_fanout_pos) {
 917                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 918
 919                         if (i != fanout_value)
 920                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 921                                              cur_fanout_pos, fanout_value, i);
 922                         cur_fanout_pos++;
 923                 }
 924
 925                 graph_commit = lookup_commit(r, &cur_oid);
 926                 if (!parse_commit_in_graph_one(g, graph_commit))
 927                         graph_report("failed to parse %s from commit-graph",
 928                                      oid_to_hex(&cur_oid));
 929         }
 930
 931         while (cur_fanout_pos < 256) {
 932                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 933
 934                 if (g->num_commits != fanout_value)
 935                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 936                                      cur_fanout_pos, fanout_value, i);
 937
 938                 cur_fanout_pos++;
 939         }
 940
 941         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
 942                 return verify_commit_graph_error;
 943
 944         for (i = 0; i < g->num_commits; i++) {
 945                 struct commit *graph_commit, *odb_commit;
 946                 struct commit_list *graph_parents, *odb_parents;
 947                 uint32_t max_generation = 0;
 948
 949                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 950
 951                 graph_commit = lookup_commit(r, &cur_oid);
 952                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
 953                 if (parse_commit_internal(odb_commit, 0, 0)) {
 954                         graph_report("failed to parse %s from object database",
 955                                      oid_to_hex(&cur_oid));
 956                         continue;
 957                 }
 958
 959                 if (oidcmp(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
 960                            get_commit_tree_oid(odb_commit)))
 961                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
 962                                      oid_to_hex(&cur_oid),
 963                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
 964                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
 965
 966                 graph_parents = graph_commit->parents;
 967                 odb_parents = odb_commit->parents;
 968
 969                 while (graph_parents) {
 970                         if (odb_parents == NULL) {
 971                                 graph_report("commit-graph parent list for commit %s is too long",
 972                                              oid_to_hex(&cur_oid));
 973                                 break;
 974                         }
 975
 976                         if (oidcmp(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
 977                                 graph_report("commit-graph parent for %s is %s != %s",
 978                                              oid_to_hex(&cur_oid),
 979                                              oid_to_hex(&graph_parents->item->object.oid),
 980                                              oid_to_hex(&odb_parents->item->object.oid));
 981
 982                         if (graph_parents->item->generation > max_generation)
 983                                 max_generation = graph_parents->item->generation;
 984
 985                         graph_parents = graph_parents->next;
 986                         odb_parents = odb_parents->next;
 987                 }
 988
 989                 if (odb_parents != NULL)
 990                         graph_report("commit-graph parent list for commit %s terminates early",
 991                                      oid_to_hex(&cur_oid));
 992
 993                 if (!graph_commit->generation) {
 994                         if (generation_zero == GENERATION_NUMBER_EXISTS)
 995                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
 996                                              oid_to_hex(&cur_oid));
 997                         generation_zero = GENERATION_ZERO_EXISTS;
 998                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
 999                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1000                                      oid_to_hex(&cur_oid));
1001
1002                 if (generation_zero == GENERATION_ZERO_EXISTS)
1003                         continue;
1004
1005                 /*
1006                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1007                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1008                  * extra logic in the following condition.
1009                  */
1010                 if (max_generation == GENERATION_NUMBER_MAX)
1011                         max_generation--;
1012
1013                 if (graph_commit->generation != max_generation + 1)
1014                         graph_report("commit-graph generation for commit %s is %u != %u",
1015                                      oid_to_hex(&cur_oid),
1016                                      graph_commit->generation,
1017                                      max_generation + 1);
1018
1019                 if (graph_commit->date != odb_commit->date)
1020                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1021                                      oid_to_hex(&cur_oid),
1022                                      graph_commit->date,
1023                                      odb_commit->date);
1024         }
1025
1026         return verify_commit_graph_error;
1027 }
1028
1029 void free_commit_graph(struct commit_graph *g)
1030 {
1031         if (!g)
1032                 return;
1033         if (g->graph_fd >= 0) {
1034                 munmap((void *)g->data, g->data_len);
1035                 g->data = NULL;
1036                 close(g->graph_fd);
1037         }
1038         free(g);
1039 }