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

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "dir.h"
   4 #include "git-compat-util.h"
   5 #include "lockfile.h"
   6 #include "pack.h"
   7 #include "packfile.h"
   8 #include "commit.h"
   9 #include "object.h"
  10 #include "refs.h"
  11 #include "revision.h"
  12 #include "sha1-lookup.h"
  13 #include "commit-graph.h"
  14 #include "object-store.h"
  15 #include "alloc.h"
  16 #include "hashmap.h"
  17 #include "replace-object.h"
  18 #include "progress.h"
  19
  20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  24 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  25
  26 #define GRAPH_DATA_WIDTH 36
  27
  28 #define GRAPH_VERSION_1 0x1
  29 #define GRAPH_VERSION GRAPH_VERSION_1
  30
  31 #define GRAPH_OID_VERSION_SHA1 1
  32 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
  33 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
  34 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
  35
  36 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
  37 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  38 #define GRAPH_PARENT_NONE 0x70000000
  39
  40 #define GRAPH_LAST_EDGE 0x80000000
  41
  42 #define GRAPH_HEADER_SIZE 8
  43 #define GRAPH_FANOUT_SIZE (4 * 256)
  44 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  45 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  46                         + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
  47
  48 char *get_commit_graph_filename(const char *obj_dir)
  49 {
  50         return xstrfmt("%s/info/commit-graph", obj_dir);
  51 }
  52
  53 static struct commit_graph *alloc_commit_graph(void)
  54 {
  55         struct commit_graph *g = xcalloc(1, sizeof(*g));
  56         g->graph_fd = -1;
  57
  58         return g;
  59 }
  60
  61 extern int read_replace_refs;
  62
  63 static int commit_graph_compatible(struct repository *r)
  64 {
  65         if (!r->gitdir)
  66                 return 0;
  67
  68         if (read_replace_refs) {
  69                 prepare_replace_object(r);
  70                 if (hashmap_get_size(&r->objects->replace_map->map))
  71                         return 0;
  72         }
  73
  74         prepare_commit_graft(r);
  75         if (r->parsed_objects && r->parsed_objects->grafts_nr)
  76                 return 0;
  77         if (is_repository_shallow(r))
  78                 return 0;
  79
  80         return 1;
  81 }
  82
  83 struct commit_graph *load_commit_graph_one(const char *graph_file)
  84 {
  85         void *graph_map;
  86         const unsigned char *data, *chunk_lookup;
  87         size_t graph_size;
  88         struct stat st;
  89         uint32_t i;
  90         struct commit_graph *graph;
  91         int fd = git_open(graph_file);
  92         uint64_t last_chunk_offset;
  93         uint32_t last_chunk_id;
  94         uint32_t graph_signature;
  95         unsigned char graph_version, hash_version;
  96
  97         if (fd < 0)
  98                 return NULL;
  99         if (fstat(fd, &st)) {
 100                 close(fd);
 101                 return NULL;
 102         }
 103         graph_size = xsize_t(st.st_size);
 104
 105         if (graph_size < GRAPH_MIN_SIZE) {
 106                 close(fd);
 107                 die(_("graph file %s is too small"), graph_file);
 108         }
 109         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
 110         data = (const unsigned char *)graph_map;
 111
 112         graph_signature = get_be32(data);
 113         if (graph_signature != GRAPH_SIGNATURE) {
 114                 error(_("graph signature %X does not match signature %X"),
 115                       graph_signature, GRAPH_SIGNATURE);
 116                 goto cleanup_fail;
 117         }
 118
 119         graph_version = *(unsigned char*)(data + 4);
 120         if (graph_version != GRAPH_VERSION) {
 121                 error(_("graph version %X does not match version %X"),
 122                       graph_version, GRAPH_VERSION);
 123                 goto cleanup_fail;
 124         }
 125
 126         hash_version = *(unsigned char*)(data + 5);
 127         if (hash_version != GRAPH_OID_VERSION) {
 128                 error(_("hash version %X does not match version %X"),
 129                       hash_version, GRAPH_OID_VERSION);
 130                 goto cleanup_fail;
 131         }
 132
 133         graph = alloc_commit_graph();
 134
 135         graph->hash_len = GRAPH_OID_LEN;
 136         graph->num_chunks = *(unsigned char*)(data + 6);
 137         graph->graph_fd = fd;
 138         graph->data = graph_map;
 139         graph->data_len = graph_size;
 140
 141         last_chunk_id = 0;
 142         last_chunk_offset = 8;
 143         chunk_lookup = data + 8;
 144         for (i = 0; i < graph->num_chunks; i++) {
 145                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 146                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 147                 int chunk_repeated = 0;
 148
 149                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 150
 151                 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
 152                         error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 153                               (uint32_t)chunk_offset);
 154                         goto cleanup_fail;
 155                 }
 156
 157                 switch (chunk_id) {
 158                 case GRAPH_CHUNKID_OIDFANOUT:
 159                         if (graph->chunk_oid_fanout)
 160                                 chunk_repeated = 1;
 161                         else
 162                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 163                         break;
 164
 165                 case GRAPH_CHUNKID_OIDLOOKUP:
 166                         if (graph->chunk_oid_lookup)
 167                                 chunk_repeated = 1;
 168                         else
 169                                 graph->chunk_oid_lookup = data + chunk_offset;
 170                         break;
 171
 172                 case GRAPH_CHUNKID_DATA:
 173                         if (graph->chunk_commit_data)
 174                                 chunk_repeated = 1;
 175                         else
 176                                 graph->chunk_commit_data = data + chunk_offset;
 177                         break;
 178
 179                 case GRAPH_CHUNKID_LARGEEDGES:
 180                         if (graph->chunk_large_edges)
 181                                 chunk_repeated = 1;
 182                         else
 183                                 graph->chunk_large_edges = data + chunk_offset;
 184                         break;
 185                 }
 186
 187                 if (chunk_repeated) {
 188                         error(_("chunk id %08x appears multiple times"), chunk_id);
 189                         goto cleanup_fail;
 190                 }
 191
 192                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 193                 {
 194                         graph->num_commits = (chunk_offset - last_chunk_offset)
 195                                              / graph->hash_len;
 196                 }
 197
 198                 last_chunk_id = chunk_id;
 199                 last_chunk_offset = chunk_offset;
 200         }
 201
 202         return graph;
 203
 204 cleanup_fail:
 205         munmap(graph_map, graph_size);
 206         close(fd);
 207         exit(1);
 208 }
 209
 210 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 211 {
 212         char *graph_name;
 213
 214         if (r->objects->commit_graph)
 215                 return;
 216
 217         graph_name = get_commit_graph_filename(obj_dir);
 218         r->objects->commit_graph =
 219                 load_commit_graph_one(graph_name);
 220
 221         FREE_AND_NULL(graph_name);
 222 }
 223
 224 /*
 225  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 226  *
 227  * On the first invocation, this function attemps to load the commit
 228  * graph if the_repository is configured to have one.
 229  */
 230 static int prepare_commit_graph(struct repository *r)
 231 {
 232         struct object_directory *odb;
 233         int config_value;
 234
 235         if (r->objects->commit_graph_attempted)
 236                 return !!r->objects->commit_graph;
 237         r->objects->commit_graph_attempted = 1;
 238
 239         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 240             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 241             !config_value))
 242                 /*
 243                  * This repository is not configured to use commit graphs, so
 244                  * do not load one. (But report commit_graph_attempted anyway
 245                  * so that commit graph loading is not attempted again for this
 246                  * repository.)
 247                  */
 248                 return 0;
 249
 250         if (!commit_graph_compatible(r))
 251                 return 0;
 252
 253         prepare_alt_odb(r);
 254         for (odb = r->objects->odb;
 255              !r->objects->commit_graph && odb;
 256              odb = odb->next)
 257                 prepare_commit_graph_one(r, odb->path);
 258         return !!r->objects->commit_graph;
 259 }
 260
 261 int generation_numbers_enabled(struct repository *r)
 262 {
 263         uint32_t first_generation;
 264         struct commit_graph *g;
 265         if (!prepare_commit_graph(r))
 266                return 0;
 267
 268         g = r->objects->commit_graph;
 269
 270         if (!g->num_commits)
 271                 return 0;
 272
 273         first_generation = get_be32(g->chunk_commit_data +
 274                                     g->hash_len + 8) >> 2;
 275
 276         return !!first_generation;
 277 }
 278
 279 void close_commit_graph(struct repository *r)
 280 {
 281         free_commit_graph(r->objects->commit_graph);
 282         r->objects->commit_graph = NULL;
 283 }
 284
 285 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 286 {
 287         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 288                             g->chunk_oid_lookup, g->hash_len, pos);
 289 }
 290
 291 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 292                                                  uint64_t pos,
 293                                                  struct commit_list **pptr)
 294 {
 295         struct commit *c;
 296         struct object_id oid;
 297
 298         if (pos >= g->num_commits)
 299                 die("invalid parent position %"PRIu64, pos);
 300
 301         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 302         c = lookup_commit(the_repository, &oid);
 303         if (!c)
 304                 die(_("could not find commit %s"), oid_to_hex(&oid));
 305         c->graph_pos = pos;
 306         return &commit_list_insert(c, pptr)->next;
 307 }
 308
 309 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 310 {
 311         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 312         item->graph_pos = pos;
 313         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 314 }
 315
 316 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 317 {
 318         uint32_t edge_value;
 319         uint32_t *parent_data_ptr;
 320         uint64_t date_low, date_high;
 321         struct commit_list **pptr;
 322         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 323
 324         item->object.parsed = 1;
 325         item->graph_pos = pos;
 326
 327         item->maybe_tree = NULL;
 328
 329         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 330         date_low = get_be32(commit_data + g->hash_len + 12);
 331         item->date = (timestamp_t)((date_high << 32) | date_low);
 332
 333         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 334
 335         pptr = &item->parents;
 336
 337         edge_value = get_be32(commit_data + g->hash_len);
 338         if (edge_value == GRAPH_PARENT_NONE)
 339                 return 1;
 340         pptr = insert_parent_or_die(g, edge_value, pptr);
 341
 342         edge_value = get_be32(commit_data + g->hash_len + 4);
 343         if (edge_value == GRAPH_PARENT_NONE)
 344                 return 1;
 345         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 346                 pptr = insert_parent_or_die(g, edge_value, pptr);
 347                 return 1;
 348         }
 349
 350         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 351                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 352         do {
 353                 edge_value = get_be32(parent_data_ptr);
 354                 pptr = insert_parent_or_die(g,
 355                                             edge_value & GRAPH_EDGE_LAST_MASK,
 356                                             pptr);
 357                 parent_data_ptr++;
 358         } while (!(edge_value & GRAPH_LAST_EDGE));
 359
 360         return 1;
 361 }
 362
 363 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 364 {
 365         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 366                 *pos = item->graph_pos;
 367                 return 1;
 368         } else {
 369                 return bsearch_graph(g, &(item->object.oid), pos);
 370         }
 371 }
 372
 373 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 374 {
 375         uint32_t pos;
 376
 377         if (item->object.parsed)
 378                 return 1;
 379
 380         if (find_commit_in_graph(item, g, &pos))
 381                 return fill_commit_in_graph(item, g, pos);
 382
 383         return 0;
 384 }
 385
 386 int parse_commit_in_graph(struct repository *r, struct commit *item)
 387 {
 388         if (!prepare_commit_graph(r))
 389                 return 0;
 390         return parse_commit_in_graph_one(r->objects->commit_graph, item);
 391 }
 392
 393 void load_commit_graph_info(struct repository *r, struct commit *item)
 394 {
 395         uint32_t pos;
 396         if (!prepare_commit_graph(r))
 397                 return;
 398         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 399                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 400 }
 401
 402 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 403 {
 404         struct object_id oid;
 405         const unsigned char *commit_data = g->chunk_commit_data +
 406                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 407
 408         hashcpy(oid.hash, commit_data);
 409         c->maybe_tree = lookup_tree(the_repository, &oid);
 410
 411         return c->maybe_tree;
 412 }
 413
 414 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 415                                                  const struct commit *c)
 416 {
 417         if (c->maybe_tree)
 418                 return c->maybe_tree;
 419         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 420                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 421
 422         return load_tree_for_commit(g, (struct commit *)c);
 423 }
 424
 425 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 426 {
 427         return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
 428 }
 429
 430 static void write_graph_chunk_fanout(struct hashfile *f,
 431                                      struct commit **commits,
 432                                      int nr_commits)
 433 {
 434         int i, count = 0;
 435         struct commit **list = commits;
 436
 437         /*
 438          * Write the first-level table (the list is sorted,
 439          * but we use a 256-entry lookup to be able to avoid
 440          * having to do eight extra binary search iterations).
 441          */
 442         for (i = 0; i < 256; i++) {
 443                 while (count < nr_commits) {
 444                         if ((*list)->object.oid.hash[0] != i)
 445                                 break;
 446                         count++;
 447                         list++;
 448                 }
 449
 450                 hashwrite_be32(f, count);
 451         }
 452 }
 453
 454 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 455                                    struct commit **commits, int nr_commits)
 456 {
 457         struct commit **list = commits;
 458         int count;
 459         for (count = 0; count < nr_commits; count++, list++)
 460                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 461 }
 462
 463 static const unsigned char *commit_to_sha1(size_t index, void *table)
 464 {
 465         struct commit **commits = table;
 466         return commits[index]->object.oid.hash;
 467 }
 468
 469 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 470                                    struct commit **commits, int nr_commits)
 471 {
 472         struct commit **list = commits;
 473         struct commit **last = commits + nr_commits;
 474         uint32_t num_extra_edges = 0;
 475
 476         while (list < last) {
 477                 struct commit_list *parent;
 478                 int edge_value;
 479                 uint32_t packedDate[2];
 480
 481                 parse_commit(*list);
 482                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 483
 484                 parent = (*list)->parents;
 485
 486                 if (!parent)
 487                         edge_value = GRAPH_PARENT_NONE;
 488                 else {
 489                         edge_value = sha1_pos(parent->item->object.oid.hash,
 490                                               commits,
 491                                               nr_commits,
 492                                               commit_to_sha1);
 493
 494                         if (edge_value < 0)
 495                                 BUG("missing parent %s for commit %s",
 496                                     oid_to_hex(&parent->item->object.oid),
 497                                     oid_to_hex(&(*list)->object.oid));
 498                 }
 499
 500                 hashwrite_be32(f, edge_value);
 501
 502                 if (parent)
 503                         parent = parent->next;
 504
 505                 if (!parent)
 506                         edge_value = GRAPH_PARENT_NONE;
 507                 else if (parent->next)
 508                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 509                 else {
 510                         edge_value = sha1_pos(parent->item->object.oid.hash,
 511                                               commits,
 512                                               nr_commits,
 513                                               commit_to_sha1);
 514                         if (edge_value < 0)
 515                                 BUG("missing parent %s for commit %s",
 516                                     oid_to_hex(&parent->item->object.oid),
 517                                     oid_to_hex(&(*list)->object.oid));
 518                 }
 519
 520                 hashwrite_be32(f, edge_value);
 521
 522                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 523                         do {
 524                                 num_extra_edges++;
 525                                 parent = parent->next;
 526                         } while (parent);
 527                 }
 528
 529                 if (sizeof((*list)->date) > 4)
 530                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 531                 else
 532                         packedDate[0] = 0;
 533
 534                 packedDate[0] |= htonl((*list)->generation << 2);
 535
 536                 packedDate[1] = htonl((*list)->date);
 537                 hashwrite(f, packedDate, 8);
 538
 539                 list++;
 540         }
 541 }
 542
 543 static void write_graph_chunk_large_edges(struct hashfile *f,
 544                                           struct commit **commits,
 545                                           int nr_commits)
 546 {
 547         struct commit **list = commits;
 548         struct commit **last = commits + nr_commits;
 549         struct commit_list *parent;
 550
 551         while (list < last) {
 552                 int num_parents = 0;
 553                 for (parent = (*list)->parents; num_parents < 3 && parent;
 554                      parent = parent->next)
 555                         num_parents++;
 556
 557                 if (num_parents <= 2) {
 558                         list++;
 559                         continue;
 560                 }
 561
 562                 /* Since num_parents > 2, this initializer is safe. */
 563                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 564                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 565                                                   commits,
 566                                                   nr_commits,
 567                                                   commit_to_sha1);
 568
 569                         if (edge_value < 0)
 570                                 BUG("missing parent %s for commit %s",
 571                                     oid_to_hex(&parent->item->object.oid),
 572                                     oid_to_hex(&(*list)->object.oid));
 573                         else if (!parent->next)
 574                                 edge_value |= GRAPH_LAST_EDGE;
 575
 576                         hashwrite_be32(f, edge_value);
 577                 }
 578
 579                 list++;
 580         }
 581 }
 582
 583 static int commit_compare(const void *_a, const void *_b)
 584 {
 585         const struct object_id *a = (const struct object_id *)_a;
 586         const struct object_id *b = (const struct object_id *)_b;
 587         return oidcmp(a, b);
 588 }
 589
 590 struct packed_commit_list {
 591         struct commit **list;
 592         int nr;
 593         int alloc;
 594 };
 595
 596 struct packed_oid_list {
 597         struct object_id *list;
 598         int nr;
 599         int alloc;
 600         struct progress *progress;
 601         int progress_done;
 602 };
 603
 604 static int add_packed_commits(const struct object_id *oid,
 605                               struct packed_git *pack,
 606                               uint32_t pos,
 607                               void *data)
 608 {
 609         struct packed_oid_list *list = (struct packed_oid_list*)data;
 610         enum object_type type;
 611         off_t offset = nth_packed_object_offset(pack, pos);
 612         struct object_info oi = OBJECT_INFO_INIT;
 613
 614         if (list->progress)
 615                 display_progress(list->progress, ++list->progress_done);
 616
 617         oi.typep = &type;
 618         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 619                 die(_("unable to get type of object %s"), oid_to_hex(oid));
 620
 621         if (type != OBJ_COMMIT)
 622                 return 0;
 623
 624         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 625         oidcpy(&(list->list[list->nr]), oid);
 626         list->nr++;
 627
 628         return 0;
 629 }
 630
 631 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 632 {
 633         struct commit_list *parent;
 634         for (parent = commit->parents; parent; parent = parent->next) {
 635                 if (!(parent->item->object.flags & UNINTERESTING)) {
 636                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 637                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 638                         oids->nr++;
 639                         parent->item->object.flags |= UNINTERESTING;
 640                 }
 641         }
 642 }
 643
 644 static void close_reachable(struct packed_oid_list *oids, int report_progress)
 645 {
 646         int i, j;
 647         struct commit *commit;
 648         struct progress *progress = NULL;
 649
 650         if (report_progress)
 651                 progress = start_delayed_progress(
 652                         _("Loading known commits in commit graph"), j = 0);
 653         for (i = 0; i < oids->nr; i++) {
 654                 display_progress(progress, ++j);
 655                 commit = lookup_commit(the_repository, &oids->list[i]);
 656                 if (commit)
 657                         commit->object.flags |= UNINTERESTING;
 658         }
 659         stop_progress(&progress);
 660
 661         /*
 662          * As this loop runs, oids->nr may grow, but not more
 663          * than the number of missing commits in the reachable
 664          * closure.
 665          */
 666         if (report_progress)
 667                 progress = start_delayed_progress(
 668                         _("Expanding reachable commits in commit graph"), j = 0);
 669         for (i = 0; i < oids->nr; i++) {
 670                 display_progress(progress, ++j);
 671                 commit = lookup_commit(the_repository, &oids->list[i]);
 672
 673                 if (commit && !parse_commit(commit))
 674                         add_missing_parents(oids, commit);
 675         }
 676         stop_progress(&progress);
 677
 678         if (report_progress)
 679                 progress = start_delayed_progress(
 680                         _("Clearing commit marks in commit graph"), j = 0);
 681         for (i = 0; i < oids->nr; i++) {
 682                 display_progress(progress, ++j);
 683                 commit = lookup_commit(the_repository, &oids->list[i]);
 684
 685                 if (commit)
 686                         commit->object.flags &= ~UNINTERESTING;
 687         }
 688         stop_progress(&progress);
 689 }
 690
 691 static void compute_generation_numbers(struct packed_commit_list* commits,
 692                                        int report_progress)
 693 {
 694         int i;
 695         struct commit_list *list = NULL;
 696         struct progress *progress = NULL;
 697
 698         if (report_progress)
 699                 progress = start_progress(
 700                         _("Computing commit graph generation numbers"),
 701                         commits->nr);
 702         for (i = 0; i < commits->nr; i++) {
 703                 display_progress(progress, i + 1);
 704                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 705                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 706                         continue;
 707
 708                 commit_list_insert(commits->list[i], &list);
 709                 while (list) {
 710                         struct commit *current = list->item;
 711                         struct commit_list *parent;
 712                         int all_parents_computed = 1;
 713                         uint32_t max_generation = 0;
 714
 715                         for (parent = current->parents; parent; parent = parent->next) {
 716                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 717                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 718                                         all_parents_computed = 0;
 719                                         commit_list_insert(parent->item, &list);
 720                                         break;
 721                                 } else if (parent->item->generation > max_generation) {
 722                                         max_generation = parent->item->generation;
 723                                 }
 724                         }
 725
 726                         if (all_parents_computed) {
 727                                 current->generation = max_generation + 1;
 728                                 pop_commit(&list);
 729
 730                                 if (current->generation > GENERATION_NUMBER_MAX)
 731                                         current->generation = GENERATION_NUMBER_MAX;
 732                         }
 733                 }
 734         }
 735         stop_progress(&progress);
 736 }
 737
 738 static int add_ref_to_list(const char *refname,
 739                            const struct object_id *oid,
 740                            int flags, void *cb_data)
 741 {
 742         struct string_list *list = (struct string_list *)cb_data;
 743
 744         string_list_append(list, oid_to_hex(oid));
 745         return 0;
 746 }
 747
 748 void write_commit_graph_reachable(const char *obj_dir, int append,
 749                                   int report_progress)
 750 {
 751         struct string_list list = STRING_LIST_INIT_DUP;
 752
 753         for_each_ref(add_ref_to_list, &list);
 754         write_commit_graph(obj_dir, NULL, &list, append, report_progress);
 755
 756         string_list_clear(&list, 0);
 757 }
 758
 759 void write_commit_graph(const char *obj_dir,
 760                         struct string_list *pack_indexes,
 761                         struct string_list *commit_hex,
 762                         int append, int report_progress)
 763 {
 764         struct packed_oid_list oids;
 765         struct packed_commit_list commits;
 766         struct hashfile *f;
 767         uint32_t i, count_distinct = 0;
 768         char *graph_name;
 769         struct lock_file lk = LOCK_INIT;
 770         uint32_t chunk_ids[5];
 771         uint64_t chunk_offsets[5];
 772         int num_chunks;
 773         int num_extra_edges;
 774         struct commit_list *parent;
 775         struct progress *progress = NULL;
 776
 777         if (!commit_graph_compatible(the_repository))
 778                 return;
 779
 780         oids.nr = 0;
 781         oids.alloc = approximate_object_count() / 32;
 782         oids.progress = NULL;
 783         oids.progress_done = 0;
 784
 785         if (append) {
 786                 prepare_commit_graph_one(the_repository, obj_dir);
 787                 if (the_repository->objects->commit_graph)
 788                         oids.alloc += the_repository->objects->commit_graph->num_commits;
 789         }
 790
 791         if (oids.alloc < 1024)
 792                 oids.alloc = 1024;
 793         ALLOC_ARRAY(oids.list, oids.alloc);
 794
 795         if (append && the_repository->objects->commit_graph) {
 796                 struct commit_graph *commit_graph =
 797                         the_repository->objects->commit_graph;
 798                 for (i = 0; i < commit_graph->num_commits; i++) {
 799                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 800                                 commit_graph->hash_len * i;
 801                         hashcpy(oids.list[oids.nr++].hash, hash);
 802                 }
 803         }
 804
 805         if (pack_indexes) {
 806                 struct strbuf packname = STRBUF_INIT;
 807                 int dirlen;
 808                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 809                 dirlen = packname.len;
 810                 if (report_progress) {
 811                         oids.progress = start_delayed_progress(
 812                                 _("Finding commits for commit graph"), 0);
 813                         oids.progress_done = 0;
 814                 }
 815                 for (i = 0; i < pack_indexes->nr; i++) {
 816                         struct packed_git *p;
 817                         strbuf_setlen(&packname, dirlen);
 818                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 819                         p = add_packed_git(packname.buf, packname.len, 1);
 820                         if (!p)
 821                                 die(_("error adding pack %s"), packname.buf);
 822                         if (open_pack_index(p))
 823                                 die(_("error opening index for %s"), packname.buf);
 824                         for_each_object_in_pack(p, add_packed_commits, &oids, 0);
 825                         close_pack(p);
 826                         free(p);
 827                 }
 828                 stop_progress(&oids.progress);
 829                 strbuf_release(&packname);
 830         }
 831
 832         if (commit_hex) {
 833                 if (report_progress)
 834                         progress = start_delayed_progress(
 835                                 _("Finding commits for commit graph"),
 836                                 commit_hex->nr);
 837                 for (i = 0; i < commit_hex->nr; i++) {
 838                         const char *end;
 839                         struct object_id oid;
 840                         struct commit *result;
 841
 842                         display_progress(progress, i + 1);
 843                         if (commit_hex->items[i].string &&
 844                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 845                                 continue;
 846
 847                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
 848
 849                         if (result) {
 850                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 851                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 852                                 oids.nr++;
 853                         }
 854                 }
 855                 stop_progress(&progress);
 856         }
 857
 858         if (!pack_indexes && !commit_hex) {
 859                 if (report_progress)
 860                         oids.progress = start_delayed_progress(
 861                                 _("Finding commits for commit graph"), 0);
 862                 for_each_packed_object(add_packed_commits, &oids, 0);
 863                 stop_progress(&oids.progress);
 864         }
 865
 866         close_reachable(&oids, report_progress);
 867
 868         QSORT(oids.list, oids.nr, commit_compare);
 869
 870         count_distinct = 1;
 871         for (i = 1; i < oids.nr; i++) {
 872                 if (!oideq(&oids.list[i - 1], &oids.list[i]))
 873                         count_distinct++;
 874         }
 875
 876         if (count_distinct >= GRAPH_EDGE_LAST_MASK)
 877                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 878
 879         commits.nr = 0;
 880         commits.alloc = count_distinct;
 881         ALLOC_ARRAY(commits.list, commits.alloc);
 882
 883         num_extra_edges = 0;
 884         for (i = 0; i < oids.nr; i++) {
 885                 int num_parents = 0;
 886                 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
 887                         continue;
 888
 889                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
 890                 parse_commit(commits.list[commits.nr]);
 891
 892                 for (parent = commits.list[commits.nr]->parents;
 893                      parent; parent = parent->next)
 894                         num_parents++;
 895
 896                 if (num_parents > 2)
 897                         num_extra_edges += num_parents - 1;
 898
 899                 commits.nr++;
 900         }
 901         num_chunks = num_extra_edges ? 4 : 3;
 902
 903         if (commits.nr >= GRAPH_EDGE_LAST_MASK)
 904                 die(_("too many commits to write graph"));
 905
 906         compute_generation_numbers(&commits, report_progress);
 907
 908         graph_name = get_commit_graph_filename(obj_dir);
 909         if (safe_create_leading_directories(graph_name)) {
 910                 UNLEAK(graph_name);
 911                 die_errno(_("unable to create leading directories of %s"),
 912                           graph_name);
 913         }
 914
 915         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 916         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 917
 918         hashwrite_be32(f, GRAPH_SIGNATURE);
 919
 920         hashwrite_u8(f, GRAPH_VERSION);
 921         hashwrite_u8(f, GRAPH_OID_VERSION);
 922         hashwrite_u8(f, num_chunks);
 923         hashwrite_u8(f, 0); /* unused padding byte */
 924
 925         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 926         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 927         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 928         if (num_extra_edges)
 929                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 930         else
 931                 chunk_ids[3] = 0;
 932         chunk_ids[4] = 0;
 933
 934         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 935         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 936         chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 937         chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 938         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 939
 940         for (i = 0; i <= num_chunks; i++) {
 941                 uint32_t chunk_write[3];
 942
 943                 chunk_write[0] = htonl(chunk_ids[i]);
 944                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 945                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 946                 hashwrite(f, chunk_write, 12);
 947         }
 948
 949         write_graph_chunk_fanout(f, commits.list, commits.nr);
 950         write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 951         write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 952         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 953
 954         close_commit_graph(the_repository);
 955         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 956         commit_lock_file(&lk);
 957
 958         free(graph_name);
 959         free(commits.list);
 960         free(oids.list);
 961 }
 962
 963 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 964 static int verify_commit_graph_error;
 965
 966 static void graph_report(const char *fmt, ...)
 967 {
 968         va_list ap;
 969
 970         verify_commit_graph_error = 1;
 971         va_start(ap, fmt);
 972         vfprintf(stderr, fmt, ap);
 973         fprintf(stderr, "\n");
 974         va_end(ap);
 975 }
 976
 977 #define GENERATION_ZERO_EXISTS 1
 978 #define GENERATION_NUMBER_EXISTS 2
 979
 980 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 981 {
 982         uint32_t i, cur_fanout_pos = 0;
 983         struct object_id prev_oid, cur_oid, checksum;
 984         int generation_zero = 0;
 985         struct hashfile *f;
 986         int devnull;
 987         struct progress *progress = NULL;
 988
 989         if (!g) {
 990                 graph_report("no commit-graph file loaded");
 991                 return 1;
 992         }
 993
 994         verify_commit_graph_error = 0;
 995
 996         if (!g->chunk_oid_fanout)
 997                 graph_report("commit-graph is missing the OID Fanout chunk");
 998         if (!g->chunk_oid_lookup)
 999                 graph_report("commit-graph is missing the OID Lookup chunk");
1000         if (!g->chunk_commit_data)
1001                 graph_report("commit-graph is missing the Commit Data chunk");
1002
1003         if (verify_commit_graph_error)
1004                 return verify_commit_graph_error;
1005
1006         devnull = open("/dev/null", O_WRONLY);
1007         f = hashfd(devnull, NULL);
1008         hashwrite(f, g->data, g->data_len - g->hash_len);
1009         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1010         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1011                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1012                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1013         }
1014
1015         for (i = 0; i < g->num_commits; i++) {
1016                 struct commit *graph_commit;
1017
1018                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1019
1020                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1021                         graph_report("commit-graph has incorrect OID order: %s then %s",
1022                                      oid_to_hex(&prev_oid),
1023                                      oid_to_hex(&cur_oid));
1024
1025                 oidcpy(&prev_oid, &cur_oid);
1026
1027                 while (cur_oid.hash[0] > cur_fanout_pos) {
1028                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1029
1030                         if (i != fanout_value)
1031                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1032                                              cur_fanout_pos, fanout_value, i);
1033                         cur_fanout_pos++;
1034                 }
1035
1036                 graph_commit = lookup_commit(r, &cur_oid);
1037                 if (!parse_commit_in_graph_one(g, graph_commit))
1038                         graph_report("failed to parse %s from commit-graph",
1039                                      oid_to_hex(&cur_oid));
1040         }
1041
1042         while (cur_fanout_pos < 256) {
1043                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1044
1045                 if (g->num_commits != fanout_value)
1046                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1047                                      cur_fanout_pos, fanout_value, i);
1048
1049                 cur_fanout_pos++;
1050         }
1051
1052         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1053                 return verify_commit_graph_error;
1054
1055         progress = start_progress(_("Verifying commits in commit graph"),
1056                                   g->num_commits);
1057         for (i = 0; i < g->num_commits; i++) {
1058                 struct commit *graph_commit, *odb_commit;
1059                 struct commit_list *graph_parents, *odb_parents;
1060                 uint32_t max_generation = 0;
1061
1062                 display_progress(progress, i + 1);
1063                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1064
1065                 graph_commit = lookup_commit(r, &cur_oid);
1066                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1067                 if (parse_commit_internal(odb_commit, 0, 0)) {
1068                         graph_report("failed to parse %s from object database",
1069                                      oid_to_hex(&cur_oid));
1070                         continue;
1071                 }
1072
1073                 if (!oideq(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
1074                            get_commit_tree_oid(odb_commit)))
1075                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1076                                      oid_to_hex(&cur_oid),
1077                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1078                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1079
1080                 graph_parents = graph_commit->parents;
1081                 odb_parents = odb_commit->parents;
1082
1083                 while (graph_parents) {
1084                         if (odb_parents == NULL) {
1085                                 graph_report("commit-graph parent list for commit %s is too long",
1086                                              oid_to_hex(&cur_oid));
1087                                 break;
1088                         }
1089
1090                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1091                                 graph_report("commit-graph parent for %s is %s != %s",
1092                                              oid_to_hex(&cur_oid),
1093                                              oid_to_hex(&graph_parents->item->object.oid),
1094                                              oid_to_hex(&odb_parents->item->object.oid));
1095
1096                         if (graph_parents->item->generation > max_generation)
1097                                 max_generation = graph_parents->item->generation;
1098
1099                         graph_parents = graph_parents->next;
1100                         odb_parents = odb_parents->next;
1101                 }
1102
1103                 if (odb_parents != NULL)
1104                         graph_report("commit-graph parent list for commit %s terminates early",
1105                                      oid_to_hex(&cur_oid));
1106
1107                 if (!graph_commit->generation) {
1108                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1109                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1110                                              oid_to_hex(&cur_oid));
1111                         generation_zero = GENERATION_ZERO_EXISTS;
1112                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1113                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1114                                      oid_to_hex(&cur_oid));
1115
1116                 if (generation_zero == GENERATION_ZERO_EXISTS)
1117                         continue;
1118
1119                 /*
1120                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1121                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1122                  * extra logic in the following condition.
1123                  */
1124                 if (max_generation == GENERATION_NUMBER_MAX)
1125                         max_generation--;
1126
1127                 if (graph_commit->generation != max_generation + 1)
1128                         graph_report("commit-graph generation for commit %s is %u != %u",
1129                                      oid_to_hex(&cur_oid),
1130                                      graph_commit->generation,
1131                                      max_generation + 1);
1132
1133                 if (graph_commit->date != odb_commit->date)
1134                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1135                                      oid_to_hex(&cur_oid),
1136                                      graph_commit->date,
1137                                      odb_commit->date);
1138         }
1139         stop_progress(&progress);
1140
1141         return verify_commit_graph_error;
1142 }
1143
1144 void free_commit_graph(struct commit_graph *g)
1145 {
1146         if (!g)
1147                 return;
1148         if (g->graph_fd >= 0) {
1149                 munmap((void *)g->data, g->data_len);
1150                 g->data = NULL;
1151                 close(g->graph_fd);
1152         }
1153         free(g);
1154 }