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

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "dir.h"
   4 #include "git-compat-util.h"
   5 #include "lockfile.h"
   6 #include "pack.h"
   7 #include "packfile.h"
   8 #include "commit.h"
   9 #include "object.h"
  10 #include "refs.h"
  11 #include "revision.h"
  12 #include "sha1-lookup.h"
  13 #include "commit-graph.h"
  14 #include "object-store.h"
  15 #include "alloc.h"
  16 #include "progress.h"
  17
  18 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  19 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  20 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  21 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  22 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  23
  24 #define GRAPH_DATA_WIDTH 36
  25
  26 #define GRAPH_VERSION_1 0x1
  27 #define GRAPH_VERSION GRAPH_VERSION_1
  28
  29 #define GRAPH_OID_VERSION_SHA1 1
  30 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
  31 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
  32 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
  33
  34 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
  35 #define GRAPH_PARENT_MISSING 0x7fffffff
  36 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  37 #define GRAPH_PARENT_NONE 0x70000000
  38
  39 #define GRAPH_LAST_EDGE 0x80000000
  40
  41 #define GRAPH_HEADER_SIZE 8
  42 #define GRAPH_FANOUT_SIZE (4 * 256)
  43 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  44 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  45                         + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
  46
  47 char *get_commit_graph_filename(const char *obj_dir)
  48 {
  49         return xstrfmt("%s/info/commit-graph", obj_dir);
  50 }
  51
  52 static struct commit_graph *alloc_commit_graph(void)
  53 {
  54         struct commit_graph *g = xcalloc(1, sizeof(*g));
  55         g->graph_fd = -1;
  56
  57         return g;
  58 }
  59
  60 struct commit_graph *load_commit_graph_one(const char *graph_file)
  61 {
  62         void *graph_map;
  63         const unsigned char *data, *chunk_lookup;
  64         size_t graph_size;
  65         struct stat st;
  66         uint32_t i;
  67         struct commit_graph *graph;
  68         int fd = git_open(graph_file);
  69         uint64_t last_chunk_offset;
  70         uint32_t last_chunk_id;
  71         uint32_t graph_signature;
  72         unsigned char graph_version, hash_version;
  73
  74         if (fd < 0)
  75                 return NULL;
  76         if (fstat(fd, &st)) {
  77                 close(fd);
  78                 return NULL;
  79         }
  80         graph_size = xsize_t(st.st_size);
  81
  82         if (graph_size < GRAPH_MIN_SIZE) {
  83                 close(fd);
  84                 die(_("graph file %s is too small"), graph_file);
  85         }
  86         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
  87         data = (const unsigned char *)graph_map;
  88
  89         graph_signature = get_be32(data);
  90         if (graph_signature != GRAPH_SIGNATURE) {
  91                 error(_("graph signature %X does not match signature %X"),
  92                       graph_signature, GRAPH_SIGNATURE);
  93                 goto cleanup_fail;
  94         }
  95
  96         graph_version = *(unsigned char*)(data + 4);
  97         if (graph_version != GRAPH_VERSION) {
  98                 error(_("graph version %X does not match version %X"),
  99                       graph_version, GRAPH_VERSION);
 100                 goto cleanup_fail;
 101         }
 102
 103         hash_version = *(unsigned char*)(data + 5);
 104         if (hash_version != GRAPH_OID_VERSION) {
 105                 error(_("hash version %X does not match version %X"),
 106                       hash_version, GRAPH_OID_VERSION);
 107                 goto cleanup_fail;
 108         }
 109
 110         graph = alloc_commit_graph();
 111
 112         graph->hash_len = GRAPH_OID_LEN;
 113         graph->num_chunks = *(unsigned char*)(data + 6);
 114         graph->graph_fd = fd;
 115         graph->data = graph_map;
 116         graph->data_len = graph_size;
 117
 118         last_chunk_id = 0;
 119         last_chunk_offset = 8;
 120         chunk_lookup = data + 8;
 121         for (i = 0; i < graph->num_chunks; i++) {
 122                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 123                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 124                 int chunk_repeated = 0;
 125
 126                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 127
 128                 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
 129                         error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 130                               (uint32_t)chunk_offset);
 131                         goto cleanup_fail;
 132                 }
 133
 134                 switch (chunk_id) {
 135                 case GRAPH_CHUNKID_OIDFANOUT:
 136                         if (graph->chunk_oid_fanout)
 137                                 chunk_repeated = 1;
 138                         else
 139                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 140                         break;
 141
 142                 case GRAPH_CHUNKID_OIDLOOKUP:
 143                         if (graph->chunk_oid_lookup)
 144                                 chunk_repeated = 1;
 145                         else
 146                                 graph->chunk_oid_lookup = data + chunk_offset;
 147                         break;
 148
 149                 case GRAPH_CHUNKID_DATA:
 150                         if (graph->chunk_commit_data)
 151                                 chunk_repeated = 1;
 152                         else
 153                                 graph->chunk_commit_data = data + chunk_offset;
 154                         break;
 155
 156                 case GRAPH_CHUNKID_LARGEEDGES:
 157                         if (graph->chunk_large_edges)
 158                                 chunk_repeated = 1;
 159                         else
 160                                 graph->chunk_large_edges = data + chunk_offset;
 161                         break;
 162                 }
 163
 164                 if (chunk_repeated) {
 165                         error(_("chunk id %08x appears multiple times"), chunk_id);
 166                         goto cleanup_fail;
 167                 }
 168
 169                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 170                 {
 171                         graph->num_commits = (chunk_offset - last_chunk_offset)
 172                                              / graph->hash_len;
 173                 }
 174
 175                 last_chunk_id = chunk_id;
 176                 last_chunk_offset = chunk_offset;
 177         }
 178
 179         return graph;
 180
 181 cleanup_fail:
 182         munmap(graph_map, graph_size);
 183         close(fd);
 184         exit(1);
 185 }
 186
 187 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 188 {
 189         char *graph_name;
 190
 191         if (r->objects->commit_graph)
 192                 return;
 193
 194         graph_name = get_commit_graph_filename(obj_dir);
 195         r->objects->commit_graph =
 196                 load_commit_graph_one(graph_name);
 197
 198         FREE_AND_NULL(graph_name);
 199 }
 200
 201 /*
 202  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 203  *
 204  * On the first invocation, this function attemps to load the commit
 205  * graph if the_repository is configured to have one.
 206  */
 207 static int prepare_commit_graph(struct repository *r)
 208 {
 209         struct alternate_object_database *alt;
 210         char *obj_dir;
 211         int config_value;
 212
 213         if (r->objects->commit_graph_attempted)
 214                 return !!r->objects->commit_graph;
 215         r->objects->commit_graph_attempted = 1;
 216
 217         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 218             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 219             !config_value))
 220                 /*
 221                  * This repository is not configured to use commit graphs, so
 222                  * do not load one. (But report commit_graph_attempted anyway
 223                  * so that commit graph loading is not attempted again for this
 224                  * repository.)
 225                  */
 226                 return 0;
 227
 228         obj_dir = r->objects->objectdir;
 229         prepare_commit_graph_one(r, obj_dir);
 230         prepare_alt_odb(r);
 231         for (alt = r->objects->alt_odb_list;
 232              !r->objects->commit_graph && alt;
 233              alt = alt->next)
 234                 prepare_commit_graph_one(r, alt->path);
 235         return !!r->objects->commit_graph;
 236 }
 237
 238 int generation_numbers_enabled(struct repository *r)
 239 {
 240         uint32_t first_generation;
 241         struct commit_graph *g;
 242         if (!prepare_commit_graph(r))
 243                return 0;
 244
 245         g = r->objects->commit_graph;
 246
 247         if (!g->num_commits)
 248                 return 0;
 249
 250         first_generation = get_be32(g->chunk_commit_data +
 251                                     g->hash_len + 8) >> 2;
 252
 253         return !!first_generation;
 254 }
 255
 256 static void close_commit_graph(void)
 257 {
 258         free_commit_graph(the_repository->objects->commit_graph);
 259         the_repository->objects->commit_graph = NULL;
 260 }
 261
 262 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 263 {
 264         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 265                             g->chunk_oid_lookup, g->hash_len, pos);
 266 }
 267
 268 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 269                                                  uint64_t pos,
 270                                                  struct commit_list **pptr)
 271 {
 272         struct commit *c;
 273         struct object_id oid;
 274
 275         if (pos >= g->num_commits)
 276                 die("invalid parent position %"PRIu64, pos);
 277
 278         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 279         c = lookup_commit(the_repository, &oid);
 280         if (!c)
 281                 die(_("could not find commit %s"), oid_to_hex(&oid));
 282         c->graph_pos = pos;
 283         return &commit_list_insert(c, pptr)->next;
 284 }
 285
 286 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 287 {
 288         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 289         item->graph_pos = pos;
 290         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 291 }
 292
 293 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 294 {
 295         uint32_t edge_value;
 296         uint32_t *parent_data_ptr;
 297         uint64_t date_low, date_high;
 298         struct commit_list **pptr;
 299         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 300
 301         item->object.parsed = 1;
 302         item->graph_pos = pos;
 303
 304         item->maybe_tree = NULL;
 305
 306         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 307         date_low = get_be32(commit_data + g->hash_len + 12);
 308         item->date = (timestamp_t)((date_high << 32) | date_low);
 309
 310         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 311
 312         pptr = &item->parents;
 313
 314         edge_value = get_be32(commit_data + g->hash_len);
 315         if (edge_value == GRAPH_PARENT_NONE)
 316                 return 1;
 317         pptr = insert_parent_or_die(g, edge_value, pptr);
 318
 319         edge_value = get_be32(commit_data + g->hash_len + 4);
 320         if (edge_value == GRAPH_PARENT_NONE)
 321                 return 1;
 322         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 323                 pptr = insert_parent_or_die(g, edge_value, pptr);
 324                 return 1;
 325         }
 326
 327         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 328                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 329         do {
 330                 edge_value = get_be32(parent_data_ptr);
 331                 pptr = insert_parent_or_die(g,
 332                                             edge_value & GRAPH_EDGE_LAST_MASK,
 333                                             pptr);
 334                 parent_data_ptr++;
 335         } while (!(edge_value & GRAPH_LAST_EDGE));
 336
 337         return 1;
 338 }
 339
 340 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 341 {
 342         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 343                 *pos = item->graph_pos;
 344                 return 1;
 345         } else {
 346                 return bsearch_graph(g, &(item->object.oid), pos);
 347         }
 348 }
 349
 350 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 351 {
 352         uint32_t pos;
 353
 354         if (item->object.parsed)
 355                 return 1;
 356
 357         if (find_commit_in_graph(item, g, &pos))
 358                 return fill_commit_in_graph(item, g, pos);
 359
 360         return 0;
 361 }
 362
 363 int parse_commit_in_graph(struct repository *r, struct commit *item)
 364 {
 365         if (!prepare_commit_graph(r))
 366                 return 0;
 367         return parse_commit_in_graph_one(r->objects->commit_graph, item);
 368 }
 369
 370 void load_commit_graph_info(struct repository *r, struct commit *item)
 371 {
 372         uint32_t pos;
 373         if (!prepare_commit_graph(r))
 374                 return;
 375         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 376                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 377 }
 378
 379 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 380 {
 381         struct object_id oid;
 382         const unsigned char *commit_data = g->chunk_commit_data +
 383                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 384
 385         hashcpy(oid.hash, commit_data);
 386         c->maybe_tree = lookup_tree(the_repository, &oid);
 387
 388         return c->maybe_tree;
 389 }
 390
 391 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 392                                                  const struct commit *c)
 393 {
 394         if (c->maybe_tree)
 395                 return c->maybe_tree;
 396         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 397                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 398
 399         return load_tree_for_commit(g, (struct commit *)c);
 400 }
 401
 402 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 403 {
 404         return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
 405 }
 406
 407 static void write_graph_chunk_fanout(struct hashfile *f,
 408                                      struct commit **commits,
 409                                      int nr_commits)
 410 {
 411         int i, count = 0;
 412         struct commit **list = commits;
 413
 414         /*
 415          * Write the first-level table (the list is sorted,
 416          * but we use a 256-entry lookup to be able to avoid
 417          * having to do eight extra binary search iterations).
 418          */
 419         for (i = 0; i < 256; i++) {
 420                 while (count < nr_commits) {
 421                         if ((*list)->object.oid.hash[0] != i)
 422                                 break;
 423                         count++;
 424                         list++;
 425                 }
 426
 427                 hashwrite_be32(f, count);
 428         }
 429 }
 430
 431 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 432                                    struct commit **commits, int nr_commits)
 433 {
 434         struct commit **list = commits;
 435         int count;
 436         for (count = 0; count < nr_commits; count++, list++)
 437                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 438 }
 439
 440 static const unsigned char *commit_to_sha1(size_t index, void *table)
 441 {
 442         struct commit **commits = table;
 443         return commits[index]->object.oid.hash;
 444 }
 445
 446 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 447                                    struct commit **commits, int nr_commits)
 448 {
 449         struct commit **list = commits;
 450         struct commit **last = commits + nr_commits;
 451         uint32_t num_extra_edges = 0;
 452
 453         while (list < last) {
 454                 struct commit_list *parent;
 455                 int edge_value;
 456                 uint32_t packedDate[2];
 457
 458                 parse_commit(*list);
 459                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 460
 461                 parent = (*list)->parents;
 462
 463                 if (!parent)
 464                         edge_value = GRAPH_PARENT_NONE;
 465                 else {
 466                         edge_value = sha1_pos(parent->item->object.oid.hash,
 467                                               commits,
 468                                               nr_commits,
 469                                               commit_to_sha1);
 470
 471                         if (edge_value < 0)
 472                                 edge_value = GRAPH_PARENT_MISSING;
 473                 }
 474
 475                 hashwrite_be32(f, edge_value);
 476
 477                 if (parent)
 478                         parent = parent->next;
 479
 480                 if (!parent)
 481                         edge_value = GRAPH_PARENT_NONE;
 482                 else if (parent->next)
 483                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 484                 else {
 485                         edge_value = sha1_pos(parent->item->object.oid.hash,
 486                                               commits,
 487                                               nr_commits,
 488                                               commit_to_sha1);
 489                         if (edge_value < 0)
 490                                 edge_value = GRAPH_PARENT_MISSING;
 491                 }
 492
 493                 hashwrite_be32(f, edge_value);
 494
 495                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 496                         do {
 497                                 num_extra_edges++;
 498                                 parent = parent->next;
 499                         } while (parent);
 500                 }
 501
 502                 if (sizeof((*list)->date) > 4)
 503                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 504                 else
 505                         packedDate[0] = 0;
 506
 507                 packedDate[0] |= htonl((*list)->generation << 2);
 508
 509                 packedDate[1] = htonl((*list)->date);
 510                 hashwrite(f, packedDate, 8);
 511
 512                 list++;
 513         }
 514 }
 515
 516 static void write_graph_chunk_large_edges(struct hashfile *f,
 517                                           struct commit **commits,
 518                                           int nr_commits)
 519 {
 520         struct commit **list = commits;
 521         struct commit **last = commits + nr_commits;
 522         struct commit_list *parent;
 523
 524         while (list < last) {
 525                 int num_parents = 0;
 526                 for (parent = (*list)->parents; num_parents < 3 && parent;
 527                      parent = parent->next)
 528                         num_parents++;
 529
 530                 if (num_parents <= 2) {
 531                         list++;
 532                         continue;
 533                 }
 534
 535                 /* Since num_parents > 2, this initializer is safe. */
 536                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 537                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 538                                                   commits,
 539                                                   nr_commits,
 540                                                   commit_to_sha1);
 541
 542                         if (edge_value < 0)
 543                                 edge_value = GRAPH_PARENT_MISSING;
 544                         else if (!parent->next)
 545                                 edge_value |= GRAPH_LAST_EDGE;
 546
 547                         hashwrite_be32(f, edge_value);
 548                 }
 549
 550                 list++;
 551         }
 552 }
 553
 554 static int commit_compare(const void *_a, const void *_b)
 555 {
 556         const struct object_id *a = (const struct object_id *)_a;
 557         const struct object_id *b = (const struct object_id *)_b;
 558         return oidcmp(a, b);
 559 }
 560
 561 struct packed_commit_list {
 562         struct commit **list;
 563         int nr;
 564         int alloc;
 565 };
 566
 567 struct packed_oid_list {
 568         struct object_id *list;
 569         int nr;
 570         int alloc;
 571         struct progress *progress;
 572         int progress_done;
 573 };
 574
 575 static int add_packed_commits(const struct object_id *oid,
 576                               struct packed_git *pack,
 577                               uint32_t pos,
 578                               void *data)
 579 {
 580         struct packed_oid_list *list = (struct packed_oid_list*)data;
 581         enum object_type type;
 582         off_t offset = nth_packed_object_offset(pack, pos);
 583         struct object_info oi = OBJECT_INFO_INIT;
 584
 585         if (list->progress)
 586                 display_progress(list->progress, ++list->progress_done);
 587
 588         oi.typep = &type;
 589         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 590                 die(_("unable to get type of object %s"), oid_to_hex(oid));
 591
 592         if (type != OBJ_COMMIT)
 593                 return 0;
 594
 595         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 596         oidcpy(&(list->list[list->nr]), oid);
 597         list->nr++;
 598
 599         return 0;
 600 }
 601
 602 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 603 {
 604         struct commit_list *parent;
 605         for (parent = commit->parents; parent; parent = parent->next) {
 606                 if (!(parent->item->object.flags & UNINTERESTING)) {
 607                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 608                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 609                         oids->nr++;
 610                         parent->item->object.flags |= UNINTERESTING;
 611                 }
 612         }
 613 }
 614
 615 static void close_reachable(struct packed_oid_list *oids, int report_progress)
 616 {
 617         int i;
 618         struct commit *commit;
 619         struct progress *progress = NULL;
 620         int j = 0;
 621
 622         if (report_progress)
 623                 progress = start_delayed_progress(
 624                         _("Annotating commits in commit graph"), 0);
 625         for (i = 0; i < oids->nr; i++) {
 626                 display_progress(progress, ++j);
 627                 commit = lookup_commit(the_repository, &oids->list[i]);
 628                 if (commit)
 629                         commit->object.flags |= UNINTERESTING;
 630         }
 631
 632         /*
 633          * As this loop runs, oids->nr may grow, but not more
 634          * than the number of missing commits in the reachable
 635          * closure.
 636          */
 637         for (i = 0; i < oids->nr; i++) {
 638                 display_progress(progress, ++j);
 639                 commit = lookup_commit(the_repository, &oids->list[i]);
 640
 641                 if (commit && !parse_commit(commit))
 642                         add_missing_parents(oids, commit);
 643         }
 644
 645         for (i = 0; i < oids->nr; i++) {
 646                 display_progress(progress, ++j);
 647                 commit = lookup_commit(the_repository, &oids->list[i]);
 648
 649                 if (commit)
 650                         commit->object.flags &= ~UNINTERESTING;
 651         }
 652         stop_progress(&progress);
 653 }
 654
 655 static void compute_generation_numbers(struct packed_commit_list* commits,
 656                                        int report_progress)
 657 {
 658         int i;
 659         struct commit_list *list = NULL;
 660         struct progress *progress = NULL;
 661
 662         if (report_progress)
 663                 progress = start_progress(
 664                         _("Computing commit graph generation numbers"),
 665                         commits->nr);
 666         for (i = 0; i < commits->nr; i++) {
 667                 display_progress(progress, i + 1);
 668                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 669                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 670                         continue;
 671
 672                 commit_list_insert(commits->list[i], &list);
 673                 while (list) {
 674                         struct commit *current = list->item;
 675                         struct commit_list *parent;
 676                         int all_parents_computed = 1;
 677                         uint32_t max_generation = 0;
 678
 679                         for (parent = current->parents; parent; parent = parent->next) {
 680                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 681                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 682                                         all_parents_computed = 0;
 683                                         commit_list_insert(parent->item, &list);
 684                                         break;
 685                                 } else if (parent->item->generation > max_generation) {
 686                                         max_generation = parent->item->generation;
 687                                 }
 688                         }
 689
 690                         if (all_parents_computed) {
 691                                 current->generation = max_generation + 1;
 692                                 pop_commit(&list);
 693
 694                                 if (current->generation > GENERATION_NUMBER_MAX)
 695                                         current->generation = GENERATION_NUMBER_MAX;
 696                         }
 697                 }
 698         }
 699         stop_progress(&progress);
 700 }
 701
 702 static int add_ref_to_list(const char *refname,
 703                            const struct object_id *oid,
 704                            int flags, void *cb_data)
 705 {
 706         struct string_list *list = (struct string_list *)cb_data;
 707
 708         string_list_append(list, oid_to_hex(oid));
 709         return 0;
 710 }
 711
 712 void write_commit_graph_reachable(const char *obj_dir, int append,
 713                                   int report_progress)
 714 {
 715         struct string_list list;
 716
 717         string_list_init(&list, 1);
 718         for_each_ref(add_ref_to_list, &list);
 719         write_commit_graph(obj_dir, NULL, &list, append, report_progress);
 720 }
 721
 722 void write_commit_graph(const char *obj_dir,
 723                         struct string_list *pack_indexes,
 724                         struct string_list *commit_hex,
 725                         int append, int report_progress)
 726 {
 727         struct packed_oid_list oids;
 728         struct packed_commit_list commits;
 729         struct hashfile *f;
 730         uint32_t i, count_distinct = 0;
 731         char *graph_name;
 732         struct lock_file lk = LOCK_INIT;
 733         uint32_t chunk_ids[5];
 734         uint64_t chunk_offsets[5];
 735         int num_chunks;
 736         int num_extra_edges;
 737         struct commit_list *parent;
 738         struct progress *progress = NULL;
 739
 740         oids.nr = 0;
 741         oids.alloc = approximate_object_count() / 4;
 742         oids.progress = NULL;
 743         oids.progress_done = 0;
 744
 745         if (append) {
 746                 prepare_commit_graph_one(the_repository, obj_dir);
 747                 if (the_repository->objects->commit_graph)
 748                         oids.alloc += the_repository->objects->commit_graph->num_commits;
 749         }
 750
 751         if (oids.alloc < 1024)
 752                 oids.alloc = 1024;
 753         ALLOC_ARRAY(oids.list, oids.alloc);
 754
 755         if (append && the_repository->objects->commit_graph) {
 756                 struct commit_graph *commit_graph =
 757                         the_repository->objects->commit_graph;
 758                 for (i = 0; i < commit_graph->num_commits; i++) {
 759                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 760                                 commit_graph->hash_len * i;
 761                         hashcpy(oids.list[oids.nr++].hash, hash);
 762                 }
 763         }
 764
 765         if (pack_indexes) {
 766                 struct strbuf packname = STRBUF_INIT;
 767                 int dirlen;
 768                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 769                 dirlen = packname.len;
 770                 if (report_progress) {
 771                         oids.progress = start_delayed_progress(
 772                                 _("Finding commits for commit graph"), 0);
 773                         oids.progress_done = 0;
 774                 }
 775                 for (i = 0; i < pack_indexes->nr; i++) {
 776                         struct packed_git *p;
 777                         strbuf_setlen(&packname, dirlen);
 778                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 779                         p = add_packed_git(packname.buf, packname.len, 1);
 780                         if (!p)
 781                                 die(_("error adding pack %s"), packname.buf);
 782                         if (open_pack_index(p))
 783                                 die(_("error opening index for %s"), packname.buf);
 784                         for_each_object_in_pack(p, add_packed_commits, &oids, 0);
 785                         close_pack(p);
 786                 }
 787                 stop_progress(&oids.progress);
 788                 strbuf_release(&packname);
 789         }
 790
 791         if (commit_hex) {
 792                 if (report_progress)
 793                         progress = start_delayed_progress(
 794                                 _("Finding commits for commit graph"),
 795                                 commit_hex->nr);
 796                 for (i = 0; i < commit_hex->nr; i++) {
 797                         const char *end;
 798                         struct object_id oid;
 799                         struct commit *result;
 800
 801                         display_progress(progress, i + 1);
 802                         if (commit_hex->items[i].string &&
 803                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 804                                 continue;
 805
 806                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
 807
 808                         if (result) {
 809                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 810                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 811                                 oids.nr++;
 812                         }
 813                 }
 814                 stop_progress(&progress);
 815         }
 816
 817         if (!pack_indexes && !commit_hex) {
 818                 if (report_progress)
 819                         oids.progress = start_delayed_progress(
 820                                 _("Finding commits for commit graph"), 0);
 821                 for_each_packed_object(add_packed_commits, &oids, 0);
 822                 stop_progress(&oids.progress);
 823         }
 824
 825         close_reachable(&oids, report_progress);
 826
 827         QSORT(oids.list, oids.nr, commit_compare);
 828
 829         count_distinct = 1;
 830         for (i = 1; i < oids.nr; i++) {
 831                 if (!oideq(&oids.list[i - 1], &oids.list[i]))
 832                         count_distinct++;
 833         }
 834
 835         if (count_distinct >= GRAPH_PARENT_MISSING)
 836                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 837
 838         commits.nr = 0;
 839         commits.alloc = count_distinct;
 840         ALLOC_ARRAY(commits.list, commits.alloc);
 841
 842         num_extra_edges = 0;
 843         for (i = 0; i < oids.nr; i++) {
 844                 int num_parents = 0;
 845                 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
 846                         continue;
 847
 848                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
 849                 parse_commit(commits.list[commits.nr]);
 850
 851                 for (parent = commits.list[commits.nr]->parents;
 852                      parent; parent = parent->next)
 853                         num_parents++;
 854
 855                 if (num_parents > 2)
 856                         num_extra_edges += num_parents - 1;
 857
 858                 commits.nr++;
 859         }
 860         num_chunks = num_extra_edges ? 4 : 3;
 861
 862         if (commits.nr >= GRAPH_PARENT_MISSING)
 863                 die(_("too many commits to write graph"));
 864
 865         compute_generation_numbers(&commits, report_progress);
 866
 867         graph_name = get_commit_graph_filename(obj_dir);
 868         if (safe_create_leading_directories(graph_name))
 869                 die_errno(_("unable to create leading directories of %s"),
 870                           graph_name);
 871
 872         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 873         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 874
 875         hashwrite_be32(f, GRAPH_SIGNATURE);
 876
 877         hashwrite_u8(f, GRAPH_VERSION);
 878         hashwrite_u8(f, GRAPH_OID_VERSION);
 879         hashwrite_u8(f, num_chunks);
 880         hashwrite_u8(f, 0); /* unused padding byte */
 881
 882         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 883         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 884         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 885         if (num_extra_edges)
 886                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 887         else
 888                 chunk_ids[3] = 0;
 889         chunk_ids[4] = 0;
 890
 891         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 892         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 893         chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 894         chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 895         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 896
 897         for (i = 0; i <= num_chunks; i++) {
 898                 uint32_t chunk_write[3];
 899
 900                 chunk_write[0] = htonl(chunk_ids[i]);
 901                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 902                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 903                 hashwrite(f, chunk_write, 12);
 904         }
 905
 906         write_graph_chunk_fanout(f, commits.list, commits.nr);
 907         write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 908         write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 909         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 910
 911         close_commit_graph();
 912         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 913         commit_lock_file(&lk);
 914
 915         free(oids.list);
 916         oids.alloc = 0;
 917         oids.nr = 0;
 918 }
 919
 920 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 921 static int verify_commit_graph_error;
 922
 923 static void graph_report(const char *fmt, ...)
 924 {
 925         va_list ap;
 926
 927         verify_commit_graph_error = 1;
 928         va_start(ap, fmt);
 929         vfprintf(stderr, fmt, ap);
 930         fprintf(stderr, "\n");
 931         va_end(ap);
 932 }
 933
 934 #define GENERATION_ZERO_EXISTS 1
 935 #define GENERATION_NUMBER_EXISTS 2
 936
 937 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 938 {
 939         uint32_t i, cur_fanout_pos = 0;
 940         struct object_id prev_oid, cur_oid, checksum;
 941         int generation_zero = 0;
 942         struct hashfile *f;
 943         int devnull;
 944         struct progress *progress = NULL;
 945
 946         if (!g) {
 947                 graph_report("no commit-graph file loaded");
 948                 return 1;
 949         }
 950
 951         verify_commit_graph_error = 0;
 952
 953         if (!g->chunk_oid_fanout)
 954                 graph_report("commit-graph is missing the OID Fanout chunk");
 955         if (!g->chunk_oid_lookup)
 956                 graph_report("commit-graph is missing the OID Lookup chunk");
 957         if (!g->chunk_commit_data)
 958                 graph_report("commit-graph is missing the Commit Data chunk");
 959
 960         if (verify_commit_graph_error)
 961                 return verify_commit_graph_error;
 962
 963         devnull = open("/dev/null", O_WRONLY);
 964         f = hashfd(devnull, NULL);
 965         hashwrite(f, g->data, g->data_len - g->hash_len);
 966         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
 967         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
 968                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
 969                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
 970         }
 971
 972         for (i = 0; i < g->num_commits; i++) {
 973                 struct commit *graph_commit;
 974
 975                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 976
 977                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
 978                         graph_report("commit-graph has incorrect OID order: %s then %s",
 979                                      oid_to_hex(&prev_oid),
 980                                      oid_to_hex(&cur_oid));
 981
 982                 oidcpy(&prev_oid, &cur_oid);
 983
 984                 while (cur_oid.hash[0] > cur_fanout_pos) {
 985                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 986
 987                         if (i != fanout_value)
 988                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 989                                              cur_fanout_pos, fanout_value, i);
 990                         cur_fanout_pos++;
 991                 }
 992
 993                 graph_commit = lookup_commit(r, &cur_oid);
 994                 if (!parse_commit_in_graph_one(g, graph_commit))
 995                         graph_report("failed to parse %s from commit-graph",
 996                                      oid_to_hex(&cur_oid));
 997         }
 998
 999         while (cur_fanout_pos < 256) {
1000                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1001
1002                 if (g->num_commits != fanout_value)
1003                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1004                                      cur_fanout_pos, fanout_value, i);
1005
1006                 cur_fanout_pos++;
1007         }
1008
1009         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1010                 return verify_commit_graph_error;
1011
1012         progress = start_progress(_("Verifying commits in commit graph"),
1013                                   g->num_commits);
1014         for (i = 0; i < g->num_commits; i++) {
1015                 struct commit *graph_commit, *odb_commit;
1016                 struct commit_list *graph_parents, *odb_parents;
1017                 uint32_t max_generation = 0;
1018
1019                 display_progress(progress, i + 1);
1020                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1021
1022                 graph_commit = lookup_commit(r, &cur_oid);
1023                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1024                 if (parse_commit_internal(odb_commit, 0, 0)) {
1025                         graph_report("failed to parse %s from object database",
1026                                      oid_to_hex(&cur_oid));
1027                         continue;
1028                 }
1029
1030                 if (!oideq(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
1031                            get_commit_tree_oid(odb_commit)))
1032                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1033                                      oid_to_hex(&cur_oid),
1034                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1035                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1036
1037                 graph_parents = graph_commit->parents;
1038                 odb_parents = odb_commit->parents;
1039
1040                 while (graph_parents) {
1041                         if (odb_parents == NULL) {
1042                                 graph_report("commit-graph parent list for commit %s is too long",
1043                                              oid_to_hex(&cur_oid));
1044                                 break;
1045                         }
1046
1047                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1048                                 graph_report("commit-graph parent for %s is %s != %s",
1049                                              oid_to_hex(&cur_oid),
1050                                              oid_to_hex(&graph_parents->item->object.oid),
1051                                              oid_to_hex(&odb_parents->item->object.oid));
1052
1053                         if (graph_parents->item->generation > max_generation)
1054                                 max_generation = graph_parents->item->generation;
1055
1056                         graph_parents = graph_parents->next;
1057                         odb_parents = odb_parents->next;
1058                 }
1059
1060                 if (odb_parents != NULL)
1061                         graph_report("commit-graph parent list for commit %s terminates early",
1062                                      oid_to_hex(&cur_oid));
1063
1064                 if (!graph_commit->generation) {
1065                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1066                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1067                                              oid_to_hex(&cur_oid));
1068                         generation_zero = GENERATION_ZERO_EXISTS;
1069                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1070                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1071                                      oid_to_hex(&cur_oid));
1072
1073                 if (generation_zero == GENERATION_ZERO_EXISTS)
1074                         continue;
1075
1076                 /*
1077                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1078                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1079                  * extra logic in the following condition.
1080                  */
1081                 if (max_generation == GENERATION_NUMBER_MAX)
1082                         max_generation--;
1083
1084                 if (graph_commit->generation != max_generation + 1)
1085                         graph_report("commit-graph generation for commit %s is %u != %u",
1086                                      oid_to_hex(&cur_oid),
1087                                      graph_commit->generation,
1088                                      max_generation + 1);
1089
1090                 if (graph_commit->date != odb_commit->date)
1091                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1092                                      oid_to_hex(&cur_oid),
1093                                      graph_commit->date,
1094                                      odb_commit->date);
1095         }
1096         stop_progress(&progress);
1097
1098         return verify_commit_graph_error;
1099 }
1100
1101 void free_commit_graph(struct commit_graph *g)
1102 {
1103         if (!g)
1104                 return;
1105         if (g->graph_fd >= 0) {
1106                 munmap((void *)g->data, g->data_len);
1107                 g->data = NULL;
1108                 close(g->graph_fd);
1109         }
1110         free(g);
1111 }