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 #include "alloc.h"
  15
  16 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  17 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  18 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  19 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  20 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  21
  22 #define GRAPH_DATA_WIDTH 36
  23
  24 #define GRAPH_VERSION_1 0x1
  25 #define GRAPH_VERSION GRAPH_VERSION_1
  26
  27 #define GRAPH_OID_VERSION_SHA1 1
  28 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
  29 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
  30 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
  31
  32 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
  33 #define GRAPH_PARENT_MISSING 0x7fffffff
  34 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  35 #define GRAPH_PARENT_NONE 0x70000000
  36
  37 #define GRAPH_LAST_EDGE 0x80000000
  38
  39 #define GRAPH_HEADER_SIZE 8
  40 #define GRAPH_FANOUT_SIZE (4 * 256)
  41 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  42 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  43                         + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
  44
  45 char *get_commit_graph_filename(const char *obj_dir)
  46 {
  47         return xstrfmt("%s/info/commit-graph", obj_dir);
  48 }
  49
  50 static struct commit_graph *alloc_commit_graph(void)
  51 {
  52         struct commit_graph *g = xcalloc(1, sizeof(*g));
  53         g->graph_fd = -1;
  54
  55         return g;
  56 }
  57
  58 struct commit_graph *load_commit_graph_one(const char *graph_file)
  59 {
  60         void *graph_map;
  61         const unsigned char *data, *chunk_lookup;
  62         size_t graph_size;
  63         struct stat st;
  64         uint32_t i;
  65         struct commit_graph *graph;
  66         int fd = git_open(graph_file);
  67         uint64_t last_chunk_offset;
  68         uint32_t last_chunk_id;
  69         uint32_t graph_signature;
  70         unsigned char graph_version, hash_version;
  71
  72         if (fd < 0)
  73                 return NULL;
  74         if (fstat(fd, &st)) {
  75                 close(fd);
  76                 return NULL;
  77         }
  78         graph_size = xsize_t(st.st_size);
  79
  80         if (graph_size < GRAPH_MIN_SIZE) {
  81                 close(fd);
  82                 die("graph file %s is too small", graph_file);
  83         }
  84         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
  85         data = (const unsigned char *)graph_map;
  86
  87         graph_signature = get_be32(data);
  88         if (graph_signature != GRAPH_SIGNATURE) {
  89                 error("graph signature %X does not match signature %X",
  90                       graph_signature, GRAPH_SIGNATURE);
  91                 goto cleanup_fail;
  92         }
  93
  94         graph_version = *(unsigned char*)(data + 4);
  95         if (graph_version != GRAPH_VERSION) {
  96                 error("graph version %X does not match version %X",
  97                       graph_version, GRAPH_VERSION);
  98                 goto cleanup_fail;
  99         }
 100
 101         hash_version = *(unsigned char*)(data + 5);
 102         if (hash_version != GRAPH_OID_VERSION) {
 103                 error("hash version %X does not match version %X",
 104                       hash_version, GRAPH_OID_VERSION);
 105                 goto cleanup_fail;
 106         }
 107
 108         graph = alloc_commit_graph();
 109
 110         graph->hash_len = GRAPH_OID_LEN;
 111         graph->num_chunks = *(unsigned char*)(data + 6);
 112         graph->graph_fd = fd;
 113         graph->data = graph_map;
 114         graph->data_len = graph_size;
 115
 116         last_chunk_id = 0;
 117         last_chunk_offset = 8;
 118         chunk_lookup = data + 8;
 119         for (i = 0; i < graph->num_chunks; i++) {
 120                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 121                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 122                 int chunk_repeated = 0;
 123
 124                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 125
 126                 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
 127                         error("improper chunk offset %08x%08x", (uint32_t)(chunk_offset >> 32),
 128                               (uint32_t)chunk_offset);
 129                         goto cleanup_fail;
 130                 }
 131
 132                 switch (chunk_id) {
 133                 case GRAPH_CHUNKID_OIDFANOUT:
 134                         if (graph->chunk_oid_fanout)
 135                                 chunk_repeated = 1;
 136                         else
 137                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 138                         break;
 139
 140                 case GRAPH_CHUNKID_OIDLOOKUP:
 141                         if (graph->chunk_oid_lookup)
 142                                 chunk_repeated = 1;
 143                         else
 144                                 graph->chunk_oid_lookup = data + chunk_offset;
 145                         break;
 146
 147                 case GRAPH_CHUNKID_DATA:
 148                         if (graph->chunk_commit_data)
 149                                 chunk_repeated = 1;
 150                         else
 151                                 graph->chunk_commit_data = data + chunk_offset;
 152                         break;
 153
 154                 case GRAPH_CHUNKID_LARGEEDGES:
 155                         if (graph->chunk_large_edges)
 156                                 chunk_repeated = 1;
 157                         else
 158                                 graph->chunk_large_edges = data + chunk_offset;
 159                         break;
 160                 }
 161
 162                 if (chunk_repeated) {
 163                         error("chunk id %08x appears multiple times", chunk_id);
 164                         goto cleanup_fail;
 165                 }
 166
 167                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 168                 {
 169                         graph->num_commits = (chunk_offset - last_chunk_offset)
 170                                              / graph->hash_len;
 171                 }
 172
 173                 last_chunk_id = chunk_id;
 174                 last_chunk_offset = chunk_offset;
 175         }
 176
 177         return graph;
 178
 179 cleanup_fail:
 180         munmap(graph_map, graph_size);
 181         close(fd);
 182         exit(1);
 183 }
 184
 185 /* global storage */
 186 static struct commit_graph *commit_graph = NULL;
 187
 188 static void prepare_commit_graph_one(const char *obj_dir)
 189 {
 190         char *graph_name;
 191
 192         if (commit_graph)
 193                 return;
 194
 195         graph_name = get_commit_graph_filename(obj_dir);
 196         commit_graph = load_commit_graph_one(graph_name);
 197
 198         FREE_AND_NULL(graph_name);
 199 }
 200
 201 static int prepare_commit_graph_run_once = 0;
 202 static void prepare_commit_graph(void)
 203 {
 204         struct alternate_object_database *alt;
 205         char *obj_dir;
 206
 207         if (prepare_commit_graph_run_once)
 208                 return;
 209         prepare_commit_graph_run_once = 1;
 210
 211         obj_dir = get_object_directory();
 212         prepare_commit_graph_one(obj_dir);
 213         prepare_alt_odb(the_repository);
 214         for (alt = the_repository->objects->alt_odb_list;
 215              !commit_graph && alt;
 216              alt = alt->next)
 217                 prepare_commit_graph_one(alt->path);
 218 }
 219
 220 static void close_commit_graph(void)
 221 {
 222         if (!commit_graph)
 223                 return;
 224
 225         if (commit_graph->graph_fd >= 0) {
 226                 munmap((void *)commit_graph->data, commit_graph->data_len);
 227                 commit_graph->data = NULL;
 228                 close(commit_graph->graph_fd);
 229         }
 230
 231         FREE_AND_NULL(commit_graph);
 232 }
 233
 234 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 235 {
 236         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 237                             g->chunk_oid_lookup, g->hash_len, pos);
 238 }
 239
 240 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 241                                                  uint64_t pos,
 242                                                  struct commit_list **pptr)
 243 {
 244         struct commit *c;
 245         struct object_id oid;
 246
 247         if (pos >= g->num_commits)
 248                 die("invalid parent position %"PRIu64, pos);
 249
 250         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 251         c = lookup_commit(&oid);
 252         if (!c)
 253                 die("could not find commit %s", oid_to_hex(&oid));
 254         c->graph_pos = pos;
 255         return &commit_list_insert(c, pptr)->next;
 256 }
 257
 258 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 259 {
 260         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 261         item->graph_pos = pos;
 262         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 263 }
 264
 265 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 266 {
 267         uint32_t edge_value;
 268         uint32_t *parent_data_ptr;
 269         uint64_t date_low, date_high;
 270         struct commit_list **pptr;
 271         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 272
 273         item->object.parsed = 1;
 274         item->graph_pos = pos;
 275
 276         item->maybe_tree = NULL;
 277
 278         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 279         date_low = get_be32(commit_data + g->hash_len + 12);
 280         item->date = (timestamp_t)((date_high << 32) | date_low);
 281
 282         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 283
 284         pptr = &item->parents;
 285
 286         edge_value = get_be32(commit_data + g->hash_len);
 287         if (edge_value == GRAPH_PARENT_NONE)
 288                 return 1;
 289         pptr = insert_parent_or_die(g, edge_value, pptr);
 290
 291         edge_value = get_be32(commit_data + g->hash_len + 4);
 292         if (edge_value == GRAPH_PARENT_NONE)
 293                 return 1;
 294         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 295                 pptr = insert_parent_or_die(g, edge_value, pptr);
 296                 return 1;
 297         }
 298
 299         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 300                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 301         do {
 302                 edge_value = get_be32(parent_data_ptr);
 303                 pptr = insert_parent_or_die(g,
 304                                             edge_value & GRAPH_EDGE_LAST_MASK,
 305                                             pptr);
 306                 parent_data_ptr++;
 307         } while (!(edge_value & GRAPH_LAST_EDGE));
 308
 309         return 1;
 310 }
 311
 312 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 313 {
 314         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 315                 *pos = item->graph_pos;
 316                 return 1;
 317         } else {
 318                 return bsearch_graph(g, &(item->object.oid), pos);
 319         }
 320 }
 321
 322 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 323 {
 324         uint32_t pos;
 325
 326         if (!core_commit_graph)
 327                 return 0;
 328         if (item->object.parsed)
 329                 return 1;
 330
 331         if (find_commit_in_graph(item, g, &pos))
 332                 return fill_commit_in_graph(item, g, pos);
 333
 334         return 0;
 335 }
 336
 337 int parse_commit_in_graph(struct commit *item)
 338 {
 339         if (!core_commit_graph)
 340                 return 0;
 341
 342         prepare_commit_graph();
 343         if (commit_graph)
 344                 return parse_commit_in_graph_one(commit_graph, item);
 345         return 0;
 346 }
 347
 348 void load_commit_graph_info(struct commit *item)
 349 {
 350         uint32_t pos;
 351         if (!core_commit_graph)
 352                 return;
 353         prepare_commit_graph();
 354         if (commit_graph && find_commit_in_graph(item, commit_graph, &pos))
 355                 fill_commit_graph_info(item, commit_graph, pos);
 356 }
 357
 358 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 359 {
 360         struct object_id oid;
 361         const unsigned char *commit_data = g->chunk_commit_data +
 362                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 363
 364         hashcpy(oid.hash, commit_data);
 365         c->maybe_tree = lookup_tree(&oid);
 366
 367         return c->maybe_tree;
 368 }
 369
 370 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 371                                                  const struct commit *c)
 372 {
 373         if (c->maybe_tree)
 374                 return c->maybe_tree;
 375         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 376                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 377
 378         return load_tree_for_commit(g, (struct commit *)c);
 379 }
 380
 381 struct tree *get_commit_tree_in_graph(const struct commit *c)
 382 {
 383         return get_commit_tree_in_graph_one(commit_graph, c);
 384 }
 385
 386 static void write_graph_chunk_fanout(struct hashfile *f,
 387                                      struct commit **commits,
 388                                      int nr_commits)
 389 {
 390         int i, count = 0;
 391         struct commit **list = commits;
 392
 393         /*
 394          * Write the first-level table (the list is sorted,
 395          * but we use a 256-entry lookup to be able to avoid
 396          * having to do eight extra binary search iterations).
 397          */
 398         for (i = 0; i < 256; i++) {
 399                 while (count < nr_commits) {
 400                         if ((*list)->object.oid.hash[0] != i)
 401                                 break;
 402                         count++;
 403                         list++;
 404                 }
 405
 406                 hashwrite_be32(f, count);
 407         }
 408 }
 409
 410 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 411                                    struct commit **commits, int nr_commits)
 412 {
 413         struct commit **list = commits;
 414         int count;
 415         for (count = 0; count < nr_commits; count++, list++)
 416                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 417 }
 418
 419 static const unsigned char *commit_to_sha1(size_t index, void *table)
 420 {
 421         struct commit **commits = table;
 422         return commits[index]->object.oid.hash;
 423 }
 424
 425 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 426                                    struct commit **commits, int nr_commits)
 427 {
 428         struct commit **list = commits;
 429         struct commit **last = commits + nr_commits;
 430         uint32_t num_extra_edges = 0;
 431
 432         while (list < last) {
 433                 struct commit_list *parent;
 434                 int edge_value;
 435                 uint32_t packedDate[2];
 436
 437                 parse_commit(*list);
 438                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 439
 440                 parent = (*list)->parents;
 441
 442                 if (!parent)
 443                         edge_value = GRAPH_PARENT_NONE;
 444                 else {
 445                         edge_value = sha1_pos(parent->item->object.oid.hash,
 446                                               commits,
 447                                               nr_commits,
 448                                               commit_to_sha1);
 449
 450                         if (edge_value < 0)
 451                                 edge_value = GRAPH_PARENT_MISSING;
 452                 }
 453
 454                 hashwrite_be32(f, edge_value);
 455
 456                 if (parent)
 457                         parent = parent->next;
 458
 459                 if (!parent)
 460                         edge_value = GRAPH_PARENT_NONE;
 461                 else if (parent->next)
 462                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 463                 else {
 464                         edge_value = sha1_pos(parent->item->object.oid.hash,
 465                                               commits,
 466                                               nr_commits,
 467                                               commit_to_sha1);
 468                         if (edge_value < 0)
 469                                 edge_value = GRAPH_PARENT_MISSING;
 470                 }
 471
 472                 hashwrite_be32(f, edge_value);
 473
 474                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 475                         do {
 476                                 num_extra_edges++;
 477                                 parent = parent->next;
 478                         } while (parent);
 479                 }
 480
 481                 if (sizeof((*list)->date) > 4)
 482                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 483                 else
 484                         packedDate[0] = 0;
 485
 486                 packedDate[0] |= htonl((*list)->generation << 2);
 487
 488                 packedDate[1] = htonl((*list)->date);
 489                 hashwrite(f, packedDate, 8);
 490
 491                 list++;
 492         }
 493 }
 494
 495 static void write_graph_chunk_large_edges(struct hashfile *f,
 496                                           struct commit **commits,
 497                                           int nr_commits)
 498 {
 499         struct commit **list = commits;
 500         struct commit **last = commits + nr_commits;
 501         struct commit_list *parent;
 502
 503         while (list < last) {
 504                 int num_parents = 0;
 505                 for (parent = (*list)->parents; num_parents < 3 && parent;
 506                      parent = parent->next)
 507                         num_parents++;
 508
 509                 if (num_parents <= 2) {
 510                         list++;
 511                         continue;
 512                 }
 513
 514                 /* Since num_parents > 2, this initializer is safe. */
 515                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 516                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 517                                                   commits,
 518                                                   nr_commits,
 519                                                   commit_to_sha1);
 520
 521                         if (edge_value < 0)
 522                                 edge_value = GRAPH_PARENT_MISSING;
 523                         else if (!parent->next)
 524                                 edge_value |= GRAPH_LAST_EDGE;
 525
 526                         hashwrite_be32(f, edge_value);
 527                 }
 528
 529                 list++;
 530         }
 531 }
 532
 533 static int commit_compare(const void *_a, const void *_b)
 534 {
 535         const struct object_id *a = (const struct object_id *)_a;
 536         const struct object_id *b = (const struct object_id *)_b;
 537         return oidcmp(a, b);
 538 }
 539
 540 struct packed_commit_list {
 541         struct commit **list;
 542         int nr;
 543         int alloc;
 544 };
 545
 546 struct packed_oid_list {
 547         struct object_id *list;
 548         int nr;
 549         int alloc;
 550 };
 551
 552 static int add_packed_commits(const struct object_id *oid,
 553                               struct packed_git *pack,
 554                               uint32_t pos,
 555                               void *data)
 556 {
 557         struct packed_oid_list *list = (struct packed_oid_list*)data;
 558         enum object_type type;
 559         off_t offset = nth_packed_object_offset(pack, pos);
 560         struct object_info oi = OBJECT_INFO_INIT;
 561
 562         oi.typep = &type;
 563         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 564                 die("unable to get type of object %s", oid_to_hex(oid));
 565
 566         if (type != OBJ_COMMIT)
 567                 return 0;
 568
 569         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 570         oidcpy(&(list->list[list->nr]), oid);
 571         list->nr++;
 572
 573         return 0;
 574 }
 575
 576 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 577 {
 578         struct commit_list *parent;
 579         for (parent = commit->parents; parent; parent = parent->next) {
 580                 if (!(parent->item->object.flags & UNINTERESTING)) {
 581                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 582                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 583                         oids->nr++;
 584                         parent->item->object.flags |= UNINTERESTING;
 585                 }
 586         }
 587 }
 588
 589 static void close_reachable(struct packed_oid_list *oids)
 590 {
 591         int i;
 592         struct commit *commit;
 593
 594         for (i = 0; i < oids->nr; i++) {
 595                 commit = lookup_commit(&oids->list[i]);
 596                 if (commit)
 597                         commit->object.flags |= UNINTERESTING;
 598         }
 599
 600         /*
 601          * As this loop runs, oids->nr may grow, but not more
 602          * than the number of missing commits in the reachable
 603          * closure.
 604          */
 605         for (i = 0; i < oids->nr; i++) {
 606                 commit = lookup_commit(&oids->list[i]);
 607
 608                 if (commit && !parse_commit(commit))
 609                         add_missing_parents(oids, commit);
 610         }
 611
 612         for (i = 0; i < oids->nr; i++) {
 613                 commit = lookup_commit(&oids->list[i]);
 614
 615                 if (commit)
 616                         commit->object.flags &= ~UNINTERESTING;
 617         }
 618 }
 619
 620 static void compute_generation_numbers(struct packed_commit_list* commits)
 621 {
 622         int i;
 623         struct commit_list *list = NULL;
 624
 625         for (i = 0; i < commits->nr; i++) {
 626                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 627                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 628                         continue;
 629
 630                 commit_list_insert(commits->list[i], &list);
 631                 while (list) {
 632                         struct commit *current = list->item;
 633                         struct commit_list *parent;
 634                         int all_parents_computed = 1;
 635                         uint32_t max_generation = 0;
 636
 637                         for (parent = current->parents; parent; parent = parent->next) {
 638                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 639                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 640                                         all_parents_computed = 0;
 641                                         commit_list_insert(parent->item, &list);
 642                                         break;
 643                                 } else if (parent->item->generation > max_generation) {
 644                                         max_generation = parent->item->generation;
 645                                 }
 646                         }
 647
 648                         if (all_parents_computed) {
 649                                 current->generation = max_generation + 1;
 650                                 pop_commit(&list);
 651
 652                                 if (current->generation > GENERATION_NUMBER_MAX)
 653                                         current->generation = GENERATION_NUMBER_MAX;
 654                         }
 655                 }
 656         }
 657 }
 658
 659 void write_commit_graph(const char *obj_dir,
 660                         struct string_list *pack_indexes,
 661                         struct string_list *commit_hex,
 662                         int append)
 663 {
 664         struct packed_oid_list oids;
 665         struct packed_commit_list commits;
 666         struct hashfile *f;
 667         uint32_t i, count_distinct = 0;
 668         char *graph_name;
 669         struct lock_file lk = LOCK_INIT;
 670         uint32_t chunk_ids[5];
 671         uint64_t chunk_offsets[5];
 672         int num_chunks;
 673         int num_extra_edges;
 674         struct commit_list *parent;
 675
 676         oids.nr = 0;
 677         oids.alloc = approximate_object_count() / 4;
 678
 679         if (append) {
 680                 prepare_commit_graph_one(obj_dir);
 681                 if (commit_graph)
 682                         oids.alloc += commit_graph->num_commits;
 683         }
 684
 685         if (oids.alloc < 1024)
 686                 oids.alloc = 1024;
 687         ALLOC_ARRAY(oids.list, oids.alloc);
 688
 689         if (append && commit_graph) {
 690                 for (i = 0; i < commit_graph->num_commits; i++) {
 691                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 692                                 commit_graph->hash_len * i;
 693                         hashcpy(oids.list[oids.nr++].hash, hash);
 694                 }
 695         }
 696
 697         if (pack_indexes) {
 698                 struct strbuf packname = STRBUF_INIT;
 699                 int dirlen;
 700                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 701                 dirlen = packname.len;
 702                 for (i = 0; i < pack_indexes->nr; i++) {
 703                         struct packed_git *p;
 704                         strbuf_setlen(&packname, dirlen);
 705                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 706                         p = add_packed_git(packname.buf, packname.len, 1);
 707                         if (!p)
 708                                 die("error adding pack %s", packname.buf);
 709                         if (open_pack_index(p))
 710                                 die("error opening index for %s", packname.buf);
 711                         for_each_object_in_pack(p, add_packed_commits, &oids);
 712                         close_pack(p);
 713                 }
 714                 strbuf_release(&packname);
 715         }
 716
 717         if (commit_hex) {
 718                 for (i = 0; i < commit_hex->nr; i++) {
 719                         const char *end;
 720                         struct object_id oid;
 721                         struct commit *result;
 722
 723                         if (commit_hex->items[i].string &&
 724                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 725                                 continue;
 726
 727                         result = lookup_commit_reference_gently(&oid, 1);
 728
 729                         if (result) {
 730                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 731                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 732                                 oids.nr++;
 733                         }
 734                 }
 735         }
 736
 737         if (!pack_indexes && !commit_hex)
 738                 for_each_packed_object(add_packed_commits, &oids, 0);
 739
 740         close_reachable(&oids);
 741
 742         QSORT(oids.list, oids.nr, commit_compare);
 743
 744         count_distinct = 1;
 745         for (i = 1; i < oids.nr; i++) {
 746                 if (oidcmp(&oids.list[i-1], &oids.list[i]))
 747                         count_distinct++;
 748         }
 749
 750         if (count_distinct >= GRAPH_PARENT_MISSING)
 751                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 752
 753         commits.nr = 0;
 754         commits.alloc = count_distinct;
 755         ALLOC_ARRAY(commits.list, commits.alloc);
 756
 757         num_extra_edges = 0;
 758         for (i = 0; i < oids.nr; i++) {
 759                 int num_parents = 0;
 760                 if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
 761                         continue;
 762
 763                 commits.list[commits.nr] = lookup_commit(&oids.list[i]);
 764                 parse_commit(commits.list[commits.nr]);
 765
 766                 for (parent = commits.list[commits.nr]->parents;
 767                      parent; parent = parent->next)
 768                         num_parents++;
 769
 770                 if (num_parents > 2)
 771                         num_extra_edges += num_parents - 1;
 772
 773                 commits.nr++;
 774         }
 775         num_chunks = num_extra_edges ? 4 : 3;
 776
 777         if (commits.nr >= GRAPH_PARENT_MISSING)
 778                 die(_("too many commits to write graph"));
 779
 780         compute_generation_numbers(&commits);
 781
 782         graph_name = get_commit_graph_filename(obj_dir);
 783         if (safe_create_leading_directories(graph_name))
 784                 die_errno(_("unable to create leading directories of %s"),
 785                           graph_name);
 786
 787         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 788         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 789
 790         hashwrite_be32(f, GRAPH_SIGNATURE);
 791
 792         hashwrite_u8(f, GRAPH_VERSION);
 793         hashwrite_u8(f, GRAPH_OID_VERSION);
 794         hashwrite_u8(f, num_chunks);
 795         hashwrite_u8(f, 0); /* unused padding byte */
 796
 797         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 798         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 799         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 800         if (num_extra_edges)
 801                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 802         else
 803                 chunk_ids[3] = 0;
 804         chunk_ids[4] = 0;
 805
 806         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 807         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 808         chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 809         chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 810         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 811
 812         for (i = 0; i <= num_chunks; i++) {
 813                 uint32_t chunk_write[3];
 814
 815                 chunk_write[0] = htonl(chunk_ids[i]);
 816                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 817                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 818                 hashwrite(f, chunk_write, 12);
 819         }
 820
 821         write_graph_chunk_fanout(f, commits.list, commits.nr);
 822         write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 823         write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 824         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 825
 826         close_commit_graph();
 827         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 828         commit_lock_file(&lk);
 829
 830         free(oids.list);
 831         oids.alloc = 0;
 832         oids.nr = 0;
 833 }
 834
 835 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 836 static int verify_commit_graph_error;
 837
 838 static void graph_report(const char *fmt, ...)
 839 {
 840         va_list ap;
 841
 842         verify_commit_graph_error = 1;
 843         va_start(ap, fmt);
 844         vfprintf(stderr, fmt, ap);
 845         fprintf(stderr, "\n");
 846         va_end(ap);
 847 }
 848
 849 #define GENERATION_ZERO_EXISTS 1
 850 #define GENERATION_NUMBER_EXISTS 2
 851
 852 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 853 {
 854         uint32_t i, cur_fanout_pos = 0;
 855         struct object_id prev_oid, cur_oid, checksum;
 856         int generation_zero = 0;
 857         struct hashfile *f;
 858         int devnull;
 859
 860         if (!g) {
 861                 graph_report("no commit-graph file loaded");
 862                 return 1;
 863         }
 864
 865         verify_commit_graph_error = 0;
 866
 867         if (!g->chunk_oid_fanout)
 868                 graph_report("commit-graph is missing the OID Fanout chunk");
 869         if (!g->chunk_oid_lookup)
 870                 graph_report("commit-graph is missing the OID Lookup chunk");
 871         if (!g->chunk_commit_data)
 872                 graph_report("commit-graph is missing the Commit Data chunk");
 873
 874         if (verify_commit_graph_error)
 875                 return verify_commit_graph_error;
 876
 877         devnull = open("/dev/null", O_WRONLY);
 878         f = hashfd(devnull, NULL);
 879         hashwrite(f, g->data, g->data_len - g->hash_len);
 880         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
 881         if (hashcmp(checksum.hash, g->data + g->data_len - g->hash_len)) {
 882                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
 883                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
 884         }
 885
 886         for (i = 0; i < g->num_commits; i++) {
 887                 struct commit *graph_commit;
 888
 889                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 890
 891                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
 892                         graph_report("commit-graph has incorrect OID order: %s then %s",
 893                                      oid_to_hex(&prev_oid),
 894                                      oid_to_hex(&cur_oid));
 895
 896                 oidcpy(&prev_oid, &cur_oid);
 897
 898                 while (cur_oid.hash[0] > cur_fanout_pos) {
 899                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 900
 901                         if (i != fanout_value)
 902                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 903                                              cur_fanout_pos, fanout_value, i);
 904                         cur_fanout_pos++;
 905                 }
 906
 907                 graph_commit = lookup_commit(&cur_oid);
 908                 if (!parse_commit_in_graph_one(g, graph_commit))
 909                         graph_report("failed to parse %s from commit-graph",
 910                                      oid_to_hex(&cur_oid));
 911         }
 912
 913         while (cur_fanout_pos < 256) {
 914                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 915
 916                 if (g->num_commits != fanout_value)
 917                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 918                                      cur_fanout_pos, fanout_value, i);
 919
 920                 cur_fanout_pos++;
 921         }
 922
 923         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
 924                 return verify_commit_graph_error;
 925
 926         for (i = 0; i < g->num_commits; i++) {
 927                 struct commit *graph_commit, *odb_commit;
 928                 struct commit_list *graph_parents, *odb_parents;
 929                 uint32_t max_generation = 0;
 930
 931                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 932
 933                 graph_commit = lookup_commit(&cur_oid);
 934                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
 935                 if (parse_commit_internal(odb_commit, 0, 0)) {
 936                         graph_report("failed to parse %s from object database",
 937                                      oid_to_hex(&cur_oid));
 938                         continue;
 939                 }
 940
 941                 if (oidcmp(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
 942                            get_commit_tree_oid(odb_commit)))
 943                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
 944                                      oid_to_hex(&cur_oid),
 945                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
 946                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
 947
 948                 graph_parents = graph_commit->parents;
 949                 odb_parents = odb_commit->parents;
 950
 951                 while (graph_parents) {
 952                         if (odb_parents == NULL) {
 953                                 graph_report("commit-graph parent list for commit %s is too long",
 954                                              oid_to_hex(&cur_oid));
 955                                 break;
 956                         }
 957
 958                         if (oidcmp(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
 959                                 graph_report("commit-graph parent for %s is %s != %s",
 960                                              oid_to_hex(&cur_oid),
 961                                              oid_to_hex(&graph_parents->item->object.oid),
 962                                              oid_to_hex(&odb_parents->item->object.oid));
 963
 964                         if (graph_parents->item->generation > max_generation)
 965                                 max_generation = graph_parents->item->generation;
 966
 967                         graph_parents = graph_parents->next;
 968                         odb_parents = odb_parents->next;
 969                 }
 970
 971                 if (odb_parents != NULL)
 972                         graph_report("commit-graph parent list for commit %s terminates early",
 973                                      oid_to_hex(&cur_oid));
 974
 975                 if (!graph_commit->generation) {
 976                         if (generation_zero == GENERATION_NUMBER_EXISTS)
 977                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
 978                                              oid_to_hex(&cur_oid));
 979                         generation_zero = GENERATION_ZERO_EXISTS;
 980                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
 981                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
 982                                      oid_to_hex(&cur_oid));
 983
 984                 if (generation_zero == GENERATION_ZERO_EXISTS)
 985                         continue;
 986
 987                 /*
 988                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
 989                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
 990                  * extra logic in the following condition.
 991                  */
 992                 if (max_generation == GENERATION_NUMBER_MAX)
 993                         max_generation--;
 994
 995                 if (graph_commit->generation != max_generation + 1)
 996                         graph_report("commit-graph generation for commit %s is %u != %u",
 997                                      oid_to_hex(&cur_oid),
 998                                      graph_commit->generation,
 999                                      max_generation + 1);
1000
1001                 if (graph_commit->date != odb_commit->date)
1002                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1003                                      oid_to_hex(&cur_oid),
1004                                      graph_commit->date,
1005                                      odb_commit->date);
1006         }
1007
1008         return verify_commit_graph_error;
1009 }