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 "revision.h"
  11 #include "sha1-lookup.h"
  12 #include "commit-graph.h"
  13 #include "object-store.h"
  14
  15 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  16 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  17 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  18 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  19 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  20
  21 #define GRAPH_DATA_WIDTH 36
  22
  23 #define GRAPH_VERSION_1 0x1
  24 #define GRAPH_VERSION GRAPH_VERSION_1
  25
  26 #define GRAPH_OID_VERSION_SHA1 1
  27 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
  28 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
  29 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
  30
  31 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
  32 #define GRAPH_PARENT_MISSING 0x7fffffff
  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 + GRAPH_OID_LEN)
  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 struct commit_graph *alloc_commit_graph(void)
  50 {
  51         struct commit_graph *g = xcalloc(1, sizeof(*g));
  52         g->graph_fd = -1;
  53
  54         return g;
  55 }
  56
  57 struct commit_graph *load_commit_graph_one(const char *graph_file)
  58 {
  59         void *graph_map;
  60         const unsigned char *data, *chunk_lookup;
  61         size_t graph_size;
  62         struct stat st;
  63         uint32_t i;
  64         struct commit_graph *graph;
  65         int fd = git_open(graph_file);
  66         uint64_t last_chunk_offset;
  67         uint32_t last_chunk_id;
  68         uint32_t graph_signature;
  69         unsigned char graph_version, hash_version;
  70
  71         if (fd < 0)
  72                 return NULL;
  73         if (fstat(fd, &st)) {
  74                 close(fd);
  75                 return NULL;
  76         }
  77         graph_size = xsize_t(st.st_size);
  78
  79         if (graph_size < GRAPH_MIN_SIZE) {
  80                 close(fd);
  81                 die("graph file %s is too small", graph_file);
  82         }
  83         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
  84         data = (const unsigned char *)graph_map;
  85
  86         graph_signature = get_be32(data);
  87         if (graph_signature != GRAPH_SIGNATURE) {
  88                 error("graph signature %X does not match signature %X",
  89                       graph_signature, GRAPH_SIGNATURE);
  90                 goto cleanup_fail;
  91         }
  92
  93         graph_version = *(unsigned char*)(data + 4);
  94         if (graph_version != GRAPH_VERSION) {
  95                 error("graph version %X does not match version %X",
  96                       graph_version, GRAPH_VERSION);
  97                 goto cleanup_fail;
  98         }
  99
 100         hash_version = *(unsigned char*)(data + 5);
 101         if (hash_version != GRAPH_OID_VERSION) {
 102                 error("hash version %X does not match version %X",
 103                       hash_version, GRAPH_OID_VERSION);
 104                 goto cleanup_fail;
 105         }
 106
 107         graph = alloc_commit_graph();
 108
 109         graph->hash_len = GRAPH_OID_LEN;
 110         graph->num_chunks = *(unsigned char*)(data + 6);
 111         graph->graph_fd = fd;
 112         graph->data = graph_map;
 113         graph->data_len = graph_size;
 114
 115         last_chunk_id = 0;
 116         last_chunk_offset = 8;
 117         chunk_lookup = data + 8;
 118         for (i = 0; i < graph->num_chunks; i++) {
 119                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 120                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 121                 int chunk_repeated = 0;
 122
 123                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 124
 125                 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
 126                         error("improper chunk offset %08x%08x", (uint32_t)(chunk_offset >> 32),
 127                               (uint32_t)chunk_offset);
 128                         goto cleanup_fail;
 129                 }
 130
 131                 switch (chunk_id) {
 132                 case GRAPH_CHUNKID_OIDFANOUT:
 133                         if (graph->chunk_oid_fanout)
 134                                 chunk_repeated = 1;
 135                         else
 136                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 137                         break;
 138
 139                 case GRAPH_CHUNKID_OIDLOOKUP:
 140                         if (graph->chunk_oid_lookup)
 141                                 chunk_repeated = 1;
 142                         else
 143                                 graph->chunk_oid_lookup = data + chunk_offset;
 144                         break;
 145
 146                 case GRAPH_CHUNKID_DATA:
 147                         if (graph->chunk_commit_data)
 148                                 chunk_repeated = 1;
 149                         else
 150                                 graph->chunk_commit_data = data + chunk_offset;
 151                         break;
 152
 153                 case GRAPH_CHUNKID_LARGEEDGES:
 154                         if (graph->chunk_large_edges)
 155                                 chunk_repeated = 1;
 156                         else
 157                                 graph->chunk_large_edges = data + chunk_offset;
 158                         break;
 159                 }
 160
 161                 if (chunk_repeated) {
 162                         error("chunk id %08x appears multiple times", chunk_id);
 163                         goto cleanup_fail;
 164                 }
 165
 166                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 167                 {
 168                         graph->num_commits = (chunk_offset - last_chunk_offset)
 169                                              / graph->hash_len;
 170                 }
 171
 172                 last_chunk_id = chunk_id;
 173                 last_chunk_offset = chunk_offset;
 174         }
 175
 176         return graph;
 177
 178 cleanup_fail:
 179         munmap(graph_map, graph_size);
 180         close(fd);
 181         exit(1);
 182 }
 183
 184 /* global storage */
 185 static struct commit_graph *commit_graph = NULL;
 186
 187 static void prepare_commit_graph_one(const char *obj_dir)
 188 {
 189         char *graph_name;
 190
 191         if (commit_graph)
 192                 return;
 193
 194         graph_name = get_commit_graph_filename(obj_dir);
 195         commit_graph = load_commit_graph_one(graph_name);
 196
 197         FREE_AND_NULL(graph_name);
 198 }
 199
 200 static int prepare_commit_graph_run_once = 0;
 201 static void prepare_commit_graph(void)
 202 {
 203         struct alternate_object_database *alt;
 204         char *obj_dir;
 205
 206         if (prepare_commit_graph_run_once)
 207                 return;
 208         prepare_commit_graph_run_once = 1;
 209
 210         obj_dir = get_object_directory();
 211         prepare_commit_graph_one(obj_dir);
 212         prepare_alt_odb(the_repository);
 213         for (alt = the_repository->objects->alt_odb_list;
 214              !commit_graph && alt;
 215              alt = alt->next)
 216                 prepare_commit_graph_one(alt->path);
 217 }
 218
 219 static void close_commit_graph(void)
 220 {
 221         if (!commit_graph)
 222                 return;
 223
 224         if (commit_graph->graph_fd >= 0) {
 225                 munmap((void *)commit_graph->data, commit_graph->data_len);
 226                 commit_graph->data = NULL;
 227                 close(commit_graph->graph_fd);
 228         }
 229
 230         FREE_AND_NULL(commit_graph);
 231 }
 232
 233 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 234 {
 235         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 236                             g->chunk_oid_lookup, g->hash_len, pos);
 237 }
 238
 239 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 240                                                  uint64_t pos,
 241                                                  struct commit_list **pptr)
 242 {
 243         struct commit *c;
 244         struct object_id oid;
 245         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 246         c = lookup_commit(&oid);
 247         if (!c)
 248                 die("could not find commit %s", oid_to_hex(&oid));
 249         c->graph_pos = pos;
 250         return &commit_list_insert(c, pptr)->next;
 251 }
 252
 253 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 254 {
 255         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 256         item->graph_pos = pos;
 257         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 258 }
 259
 260 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 261 {
 262         uint32_t edge_value;
 263         uint32_t *parent_data_ptr;
 264         uint64_t date_low, date_high;
 265         struct commit_list **pptr;
 266         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 267
 268         item->object.parsed = 1;
 269         item->graph_pos = pos;
 270
 271         item->maybe_tree = NULL;
 272
 273         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 274         date_low = get_be32(commit_data + g->hash_len + 12);
 275         item->date = (timestamp_t)((date_high << 32) | date_low);
 276
 277         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 278
 279         pptr = &item->parents;
 280
 281         edge_value = get_be32(commit_data + g->hash_len);
 282         if (edge_value == GRAPH_PARENT_NONE)
 283                 return 1;
 284         pptr = insert_parent_or_die(g, edge_value, pptr);
 285
 286         edge_value = get_be32(commit_data + g->hash_len + 4);
 287         if (edge_value == GRAPH_PARENT_NONE)
 288                 return 1;
 289         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 290                 pptr = insert_parent_or_die(g, edge_value, pptr);
 291                 return 1;
 292         }
 293
 294         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 295                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 296         do {
 297                 edge_value = get_be32(parent_data_ptr);
 298                 pptr = insert_parent_or_die(g,
 299                                             edge_value & GRAPH_EDGE_LAST_MASK,
 300                                             pptr);
 301                 parent_data_ptr++;
 302         } while (!(edge_value & GRAPH_LAST_EDGE));
 303
 304         return 1;
 305 }
 306
 307 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 308 {
 309         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 310                 *pos = item->graph_pos;
 311                 return 1;
 312         } else {
 313                 return bsearch_graph(g, &(item->object.oid), pos);
 314         }
 315 }
 316
 317 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 318 {
 319         uint32_t pos;
 320
 321         if (!core_commit_graph)
 322                 return 0;
 323         if (item->object.parsed)
 324                 return 1;
 325
 326         if (find_commit_in_graph(item, g, &pos))
 327                 return fill_commit_in_graph(item, g, pos);
 328
 329         return 0;
 330 }
 331
 332 int parse_commit_in_graph(struct commit *item)
 333 {
 334         if (!core_commit_graph)
 335                 return 0;
 336
 337         prepare_commit_graph();
 338         if (commit_graph)
 339                 return parse_commit_in_graph_one(commit_graph, item);
 340         return 0;
 341 }
 342
 343 void load_commit_graph_info(struct commit *item)
 344 {
 345         uint32_t pos;
 346         if (!core_commit_graph)
 347                 return;
 348         prepare_commit_graph();
 349         if (commit_graph && find_commit_in_graph(item, commit_graph, &pos))
 350                 fill_commit_graph_info(item, commit_graph, pos);
 351 }
 352
 353 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 354 {
 355         struct object_id oid;
 356         const unsigned char *commit_data = g->chunk_commit_data +
 357                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 358
 359         hashcpy(oid.hash, commit_data);
 360         c->maybe_tree = lookup_tree(&oid);
 361
 362         return c->maybe_tree;
 363 }
 364
 365 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 366                                                  const struct commit *c)
 367 {
 368         if (c->maybe_tree)
 369                 return c->maybe_tree;
 370         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 371                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 372
 373         return load_tree_for_commit(g, (struct commit *)c);
 374 }
 375
 376 struct tree *get_commit_tree_in_graph(const struct commit *c)
 377 {
 378         return get_commit_tree_in_graph_one(commit_graph, c);
 379 }
 380
 381 static void write_graph_chunk_fanout(struct hashfile *f,
 382                                      struct commit **commits,
 383                                      int nr_commits)
 384 {
 385         int i, count = 0;
 386         struct commit **list = commits;
 387
 388         /*
 389          * Write the first-level table (the list is sorted,
 390          * but we use a 256-entry lookup to be able to avoid
 391          * having to do eight extra binary search iterations).
 392          */
 393         for (i = 0; i < 256; i++) {
 394                 while (count < nr_commits) {
 395                         if ((*list)->object.oid.hash[0] != i)
 396                                 break;
 397                         count++;
 398                         list++;
 399                 }
 400
 401                 hashwrite_be32(f, count);
 402         }
 403 }
 404
 405 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 406                                    struct commit **commits, int nr_commits)
 407 {
 408         struct commit **list = commits;
 409         int count;
 410         for (count = 0; count < nr_commits; count++, list++)
 411                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 412 }
 413
 414 static const unsigned char *commit_to_sha1(size_t index, void *table)
 415 {
 416         struct commit **commits = table;
 417         return commits[index]->object.oid.hash;
 418 }
 419
 420 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 421                                    struct commit **commits, int nr_commits)
 422 {
 423         struct commit **list = commits;
 424         struct commit **last = commits + nr_commits;
 425         uint32_t num_extra_edges = 0;
 426
 427         while (list < last) {
 428                 struct commit_list *parent;
 429                 int edge_value;
 430                 uint32_t packedDate[2];
 431
 432                 parse_commit(*list);
 433                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 434
 435                 parent = (*list)->parents;
 436
 437                 if (!parent)
 438                         edge_value = GRAPH_PARENT_NONE;
 439                 else {
 440                         edge_value = sha1_pos(parent->item->object.oid.hash,
 441                                               commits,
 442                                               nr_commits,
 443                                               commit_to_sha1);
 444
 445                         if (edge_value < 0)
 446                                 edge_value = GRAPH_PARENT_MISSING;
 447                 }
 448
 449                 hashwrite_be32(f, edge_value);
 450
 451                 if (parent)
 452                         parent = parent->next;
 453
 454                 if (!parent)
 455                         edge_value = GRAPH_PARENT_NONE;
 456                 else if (parent->next)
 457                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 458                 else {
 459                         edge_value = sha1_pos(parent->item->object.oid.hash,
 460                                               commits,
 461                                               nr_commits,
 462                                               commit_to_sha1);
 463                         if (edge_value < 0)
 464                                 edge_value = GRAPH_PARENT_MISSING;
 465                 }
 466
 467                 hashwrite_be32(f, edge_value);
 468
 469                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 470                         do {
 471                                 num_extra_edges++;
 472                                 parent = parent->next;
 473                         } while (parent);
 474                 }
 475
 476                 if (sizeof((*list)->date) > 4)
 477                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 478                 else
 479                         packedDate[0] = 0;
 480
 481                 packedDate[0] |= htonl((*list)->generation << 2);
 482
 483                 packedDate[1] = htonl((*list)->date);
 484                 hashwrite(f, packedDate, 8);
 485
 486                 list++;
 487         }
 488 }
 489
 490 static void write_graph_chunk_large_edges(struct hashfile *f,
 491                                           struct commit **commits,
 492                                           int nr_commits)
 493 {
 494         struct commit **list = commits;
 495         struct commit **last = commits + nr_commits;
 496         struct commit_list *parent;
 497
 498         while (list < last) {
 499                 int num_parents = 0;
 500                 for (parent = (*list)->parents; num_parents < 3 && parent;
 501                      parent = parent->next)
 502                         num_parents++;
 503
 504                 if (num_parents <= 2) {
 505                         list++;
 506                         continue;
 507                 }
 508
 509                 /* Since num_parents > 2, this initializer is safe. */
 510                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 511                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 512                                                   commits,
 513                                                   nr_commits,
 514                                                   commit_to_sha1);
 515
 516                         if (edge_value < 0)
 517                                 edge_value = GRAPH_PARENT_MISSING;
 518                         else if (!parent->next)
 519                                 edge_value |= GRAPH_LAST_EDGE;
 520
 521                         hashwrite_be32(f, edge_value);
 522                 }
 523
 524                 list++;
 525         }
 526 }
 527
 528 static int commit_compare(const void *_a, const void *_b)
 529 {
 530         const struct object_id *a = (const struct object_id *)_a;
 531         const struct object_id *b = (const struct object_id *)_b;
 532         return oidcmp(a, b);
 533 }
 534
 535 struct packed_commit_list {
 536         struct commit **list;
 537         int nr;
 538         int alloc;
 539 };
 540
 541 struct packed_oid_list {
 542         struct object_id *list;
 543         int nr;
 544         int alloc;
 545 };
 546
 547 static int add_packed_commits(const struct object_id *oid,
 548                               struct packed_git *pack,
 549                               uint32_t pos,
 550                               void *data)
 551 {
 552         struct packed_oid_list *list = (struct packed_oid_list*)data;
 553         enum object_type type;
 554         off_t offset = nth_packed_object_offset(pack, pos);
 555         struct object_info oi = OBJECT_INFO_INIT;
 556
 557         oi.typep = &type;
 558         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 559                 die("unable to get type of object %s", oid_to_hex(oid));
 560
 561         if (type != OBJ_COMMIT)
 562                 return 0;
 563
 564         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 565         oidcpy(&(list->list[list->nr]), oid);
 566         list->nr++;
 567
 568         return 0;
 569 }
 570
 571 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 572 {
 573         struct commit_list *parent;
 574         for (parent = commit->parents; parent; parent = parent->next) {
 575                 if (!(parent->item->object.flags & UNINTERESTING)) {
 576                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 577                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 578                         oids->nr++;
 579                         parent->item->object.flags |= UNINTERESTING;
 580                 }
 581         }
 582 }
 583
 584 static void close_reachable(struct packed_oid_list *oids)
 585 {
 586         int i;
 587         struct commit *commit;
 588
 589         for (i = 0; i < oids->nr; i++) {
 590                 commit = lookup_commit(&oids->list[i]);
 591                 if (commit)
 592                         commit->object.flags |= UNINTERESTING;
 593         }
 594
 595         /*
 596          * As this loop runs, oids->nr may grow, but not more
 597          * than the number of missing commits in the reachable
 598          * closure.
 599          */
 600         for (i = 0; i < oids->nr; i++) {
 601                 commit = lookup_commit(&oids->list[i]);
 602
 603                 if (commit && !parse_commit(commit))
 604                         add_missing_parents(oids, commit);
 605         }
 606
 607         for (i = 0; i < oids->nr; i++) {
 608                 commit = lookup_commit(&oids->list[i]);
 609
 610                 if (commit)
 611                         commit->object.flags &= ~UNINTERESTING;
 612         }
 613 }
 614
 615 static void compute_generation_numbers(struct packed_commit_list* commits)
 616 {
 617         int i;
 618         struct commit_list *list = NULL;
 619
 620         for (i = 0; i < commits->nr; i++) {
 621                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 622                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 623                         continue;
 624
 625                 commit_list_insert(commits->list[i], &list);
 626                 while (list) {
 627                         struct commit *current = list->item;
 628                         struct commit_list *parent;
 629                         int all_parents_computed = 1;
 630                         uint32_t max_generation = 0;
 631
 632                         for (parent = current->parents; parent; parent = parent->next) {
 633                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 634                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 635                                         all_parents_computed = 0;
 636                                         commit_list_insert(parent->item, &list);
 637                                         break;
 638                                 } else if (parent->item->generation > max_generation) {
 639                                         max_generation = parent->item->generation;
 640                                 }
 641                         }
 642
 643                         if (all_parents_computed) {
 644                                 current->generation = max_generation + 1;
 645                                 pop_commit(&list);
 646
 647                                 if (current->generation > GENERATION_NUMBER_MAX)
 648                                         current->generation = GENERATION_NUMBER_MAX;
 649                         }
 650                 }
 651         }
 652 }
 653
 654 void write_commit_graph(const char *obj_dir,
 655                         const char **pack_indexes,
 656                         int nr_packs,
 657                         const char **commit_hex,
 658                         int nr_commits,
 659                         int append)
 660 {
 661         struct packed_oid_list oids;
 662         struct packed_commit_list commits;
 663         struct hashfile *f;
 664         uint32_t i, count_distinct = 0;
 665         char *graph_name;
 666         struct lock_file lk = LOCK_INIT;
 667         uint32_t chunk_ids[5];
 668         uint64_t chunk_offsets[5];
 669         int num_chunks;
 670         int num_extra_edges;
 671         struct commit_list *parent;
 672
 673         oids.nr = 0;
 674         oids.alloc = approximate_object_count() / 4;
 675
 676         if (append) {
 677                 prepare_commit_graph_one(obj_dir);
 678                 if (commit_graph)
 679                         oids.alloc += commit_graph->num_commits;
 680         }
 681
 682         if (oids.alloc < 1024)
 683                 oids.alloc = 1024;
 684         ALLOC_ARRAY(oids.list, oids.alloc);
 685
 686         if (append && commit_graph) {
 687                 for (i = 0; i < commit_graph->num_commits; i++) {
 688                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 689                                 commit_graph->hash_len * i;
 690                         hashcpy(oids.list[oids.nr++].hash, hash);
 691                 }
 692         }
 693
 694         if (pack_indexes) {
 695                 struct strbuf packname = STRBUF_INIT;
 696                 int dirlen;
 697                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 698                 dirlen = packname.len;
 699                 for (i = 0; i < nr_packs; i++) {
 700                         struct packed_git *p;
 701                         strbuf_setlen(&packname, dirlen);
 702                         strbuf_addstr(&packname, pack_indexes[i]);
 703                         p = add_packed_git(packname.buf, packname.len, 1);
 704                         if (!p)
 705                                 die("error adding pack %s", packname.buf);
 706                         if (open_pack_index(p))
 707                                 die("error opening index for %s", packname.buf);
 708                         for_each_object_in_pack(p, add_packed_commits, &oids);
 709                         close_pack(p);
 710                 }
 711                 strbuf_release(&packname);
 712         }
 713
 714         if (commit_hex) {
 715                 for (i = 0; i < nr_commits; i++) {
 716                         const char *end;
 717                         struct object_id oid;
 718                         struct commit *result;
 719
 720                         if (commit_hex[i] && parse_oid_hex(commit_hex[i], &oid, &end))
 721                                 continue;
 722
 723                         result = lookup_commit_reference_gently(&oid, 1);
 724
 725                         if (result) {
 726                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 727                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 728                                 oids.nr++;
 729                         }
 730                 }
 731         }
 732
 733         if (!pack_indexes && !commit_hex)
 734                 for_each_packed_object(add_packed_commits, &oids, 0);
 735
 736         close_reachable(&oids);
 737
 738         QSORT(oids.list, oids.nr, commit_compare);
 739
 740         count_distinct = 1;
 741         for (i = 1; i < oids.nr; i++) {
 742                 if (oidcmp(&oids.list[i-1], &oids.list[i]))
 743                         count_distinct++;
 744         }
 745
 746         if (count_distinct >= GRAPH_PARENT_MISSING)
 747                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 748
 749         commits.nr = 0;
 750         commits.alloc = count_distinct;
 751         ALLOC_ARRAY(commits.list, commits.alloc);
 752
 753         num_extra_edges = 0;
 754         for (i = 0; i < oids.nr; i++) {
 755                 int num_parents = 0;
 756                 if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
 757                         continue;
 758
 759                 commits.list[commits.nr] = lookup_commit(&oids.list[i]);
 760                 parse_commit(commits.list[commits.nr]);
 761
 762                 for (parent = commits.list[commits.nr]->parents;
 763                      parent; parent = parent->next)
 764                         num_parents++;
 765
 766                 if (num_parents > 2)
 767                         num_extra_edges += num_parents - 1;
 768
 769                 commits.nr++;
 770         }
 771         num_chunks = num_extra_edges ? 4 : 3;
 772
 773         if (commits.nr >= GRAPH_PARENT_MISSING)
 774                 die(_("too many commits to write graph"));
 775
 776         compute_generation_numbers(&commits);
 777
 778         graph_name = get_commit_graph_filename(obj_dir);
 779         if (safe_create_leading_directories(graph_name))
 780                 die_errno(_("unable to create leading directories of %s"),
 781                           graph_name);
 782
 783         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 784         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 785
 786         hashwrite_be32(f, GRAPH_SIGNATURE);
 787
 788         hashwrite_u8(f, GRAPH_VERSION);
 789         hashwrite_u8(f, GRAPH_OID_VERSION);
 790         hashwrite_u8(f, num_chunks);
 791         hashwrite_u8(f, 0); /* unused padding byte */
 792
 793         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 794         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 795         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 796         if (num_extra_edges)
 797                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 798         else
 799                 chunk_ids[3] = 0;
 800         chunk_ids[4] = 0;
 801
 802         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 803         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 804         chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 805         chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 806         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 807
 808         for (i = 0; i <= num_chunks; i++) {
 809                 uint32_t chunk_write[3];
 810
 811                 chunk_write[0] = htonl(chunk_ids[i]);
 812                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 813                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 814                 hashwrite(f, chunk_write, 12);
 815         }
 816
 817         write_graph_chunk_fanout(f, commits.list, commits.nr);
 818         write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 819         write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 820         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 821
 822         close_commit_graph();
 823         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 824         commit_lock_file(&lk);
 825
 826         free(oids.list);
 827         oids.alloc = 0;
 828         oids.nr = 0;
 829 }
 830
 831 static int verify_commit_graph_error;
 832
 833 static void graph_report(const char *fmt, ...)
 834 {
 835         va_list ap;
 836
 837         verify_commit_graph_error = 1;
 838         va_start(ap, fmt);
 839         vfprintf(stderr, fmt, ap);
 840         fprintf(stderr, "\n");
 841         va_end(ap);
 842 }
 843
 844 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 845 {
 846         uint32_t i, cur_fanout_pos = 0;
 847         struct object_id prev_oid, cur_oid;
 848
 849         if (!g) {
 850                 graph_report("no commit-graph file loaded");
 851                 return 1;
 852         }
 853
 854         verify_commit_graph_error = 0;
 855
 856         if (!g->chunk_oid_fanout)
 857                 graph_report("commit-graph is missing the OID Fanout chunk");
 858         if (!g->chunk_oid_lookup)
 859                 graph_report("commit-graph is missing the OID Lookup chunk");
 860         if (!g->chunk_commit_data)
 861                 graph_report("commit-graph is missing the Commit Data chunk");
 862
 863         if (verify_commit_graph_error)
 864                 return verify_commit_graph_error;
 865
 866         for (i = 0; i < g->num_commits; i++) {
 867                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 868
 869                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
 870                         graph_report("commit-graph has incorrect OID order: %s then %s",
 871                                      oid_to_hex(&prev_oid),
 872                                      oid_to_hex(&cur_oid));
 873
 874                 oidcpy(&prev_oid, &cur_oid);
 875
 876                 while (cur_oid.hash[0] > cur_fanout_pos) {
 877                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 878
 879                         if (i != fanout_value)
 880                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 881                                              cur_fanout_pos, fanout_value, i);
 882                         cur_fanout_pos++;
 883                 }
 884         }
 885
 886         while (cur_fanout_pos < 256) {
 887                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 888
 889                 if (g->num_commits != fanout_value)
 890                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 891                                      cur_fanout_pos, fanout_value, i);
 892
 893                 cur_fanout_pos++;
 894         }
 895
 896         return verify_commit_graph_error;
 897 }