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_EXTRAEDGES 0x45444745 /* "EDGE" */
  25
  26 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
  27
  28 #define GRAPH_VERSION_1 0x1
  29 #define GRAPH_VERSION GRAPH_VERSION_1
  30
  31 #define GRAPH_EXTRA_EDGES_NEEDED 0x80000000
  32 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  33 #define GRAPH_PARENT_NONE 0x70000000
  34
  35 #define GRAPH_LAST_EDGE 0x80000000
  36
  37 #define GRAPH_HEADER_SIZE 8
  38 #define GRAPH_FANOUT_SIZE (4 * 256)
  39 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  40 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  41                         + GRAPH_FANOUT_SIZE + the_hash_algo->rawsz)
  42
  43 char *get_commit_graph_filename(const char *obj_dir)
  44 {
  45         return xstrfmt("%s/info/commit-graph", obj_dir);
  46 }
  47
  48 static uint8_t oid_version(void)
  49 {
  50         return 1;
  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 int open_commit_graph(const char *graph_file, int *fd, struct stat *st)
  84 {
  85         *fd = git_open(graph_file);
  86         if (*fd < 0)
  87                 return 0;
  88         if (fstat(*fd, st)) {
  89                 close(*fd);
  90                 return 0;
  91         }
  92         return 1;
  93 }
  94
  95 struct commit_graph *load_commit_graph_one_fd_st(const char *graph_file,
  96                                                  int fd, struct stat *st)
  97 {
  98         void *graph_map;
  99         size_t graph_size;
 100         struct commit_graph *ret;
 101
 102         graph_size = xsize_t(st->st_size);
 103
 104         if (graph_size < GRAPH_MIN_SIZE) {
 105                 close(fd);
 106                 error(_("graph file %s is too small"), graph_file);
 107                 return NULL;
 108         }
 109         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
 110         ret = parse_commit_graph(graph_map, fd, graph_size);
 111
 112         if (!ret) {
 113                 munmap(graph_map, graph_size);
 114                 close(fd);
 115         }
 116
 117         return ret;
 118 }
 119
 120 static int verify_commit_graph_lite(struct commit_graph *g)
 121 {
 122         /*
 123          * Basic validation shared between parse_commit_graph()
 124          * which'll be called every time the graph is used, and the
 125          * much more expensive verify_commit_graph() used by
 126          * "commit-graph verify".
 127          *
 128          * There should only be very basic checks here to ensure that
 129          * we don't e.g. segfault in fill_commit_in_graph(), but
 130          * because this is a very hot codepath nothing that e.g. loops
 131          * over g->num_commits, or runs a checksum on the commit-graph
 132          * itself.
 133          */
 134         if (!g->chunk_oid_fanout) {
 135                 error("commit-graph is missing the OID Fanout chunk");
 136                 return 1;
 137         }
 138         if (!g->chunk_oid_lookup) {
 139                 error("commit-graph is missing the OID Lookup chunk");
 140                 return 1;
 141         }
 142         if (!g->chunk_commit_data) {
 143                 error("commit-graph is missing the Commit Data chunk");
 144                 return 1;
 145         }
 146
 147         return 0;
 148 }
 149
 150 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
 151                                         size_t graph_size)
 152 {
 153         const unsigned char *data, *chunk_lookup;
 154         uint32_t i;
 155         struct commit_graph *graph;
 156         uint64_t last_chunk_offset;
 157         uint32_t last_chunk_id;
 158         uint32_t graph_signature;
 159         unsigned char graph_version, hash_version;
 160
 161         if (!graph_map)
 162                 return NULL;
 163
 164         if (graph_size < GRAPH_MIN_SIZE)
 165                 return NULL;
 166
 167         data = (const unsigned char *)graph_map;
 168
 169         graph_signature = get_be32(data);
 170         if (graph_signature != GRAPH_SIGNATURE) {
 171                 error(_("graph signature %X does not match signature %X"),
 172                       graph_signature, GRAPH_SIGNATURE);
 173                 return NULL;
 174         }
 175
 176         graph_version = *(unsigned char*)(data + 4);
 177         if (graph_version != GRAPH_VERSION) {
 178                 error(_("graph version %X does not match version %X"),
 179                       graph_version, GRAPH_VERSION);
 180                 return NULL;
 181         }
 182
 183         hash_version = *(unsigned char*)(data + 5);
 184         if (hash_version != oid_version()) {
 185                 error(_("hash version %X does not match version %X"),
 186                       hash_version, oid_version());
 187                 return NULL;
 188         }
 189
 190         graph = alloc_commit_graph();
 191
 192         graph->hash_len = the_hash_algo->rawsz;
 193         graph->num_chunks = *(unsigned char*)(data + 6);
 194         graph->graph_fd = fd;
 195         graph->data = graph_map;
 196         graph->data_len = graph_size;
 197
 198         last_chunk_id = 0;
 199         last_chunk_offset = 8;
 200         chunk_lookup = data + 8;
 201         for (i = 0; i < graph->num_chunks; i++) {
 202                 uint32_t chunk_id;
 203                 uint64_t chunk_offset;
 204                 int chunk_repeated = 0;
 205
 206                 if (data + graph_size - chunk_lookup <
 207                     GRAPH_CHUNKLOOKUP_WIDTH) {
 208                         error(_("chunk lookup table entry missing; graph file may be incomplete"));
 209                         free(graph);
 210                         return NULL;
 211                 }
 212
 213                 chunk_id = get_be32(chunk_lookup + 0);
 214                 chunk_offset = get_be64(chunk_lookup + 4);
 215
 216                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 217
 218                 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
 219                         error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 220                               (uint32_t)chunk_offset);
 221                         free(graph);
 222                         return NULL;
 223                 }
 224
 225                 switch (chunk_id) {
 226                 case GRAPH_CHUNKID_OIDFANOUT:
 227                         if (graph->chunk_oid_fanout)
 228                                 chunk_repeated = 1;
 229                         else
 230                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 231                         break;
 232
 233                 case GRAPH_CHUNKID_OIDLOOKUP:
 234                         if (graph->chunk_oid_lookup)
 235                                 chunk_repeated = 1;
 236                         else
 237                                 graph->chunk_oid_lookup = data + chunk_offset;
 238                         break;
 239
 240                 case GRAPH_CHUNKID_DATA:
 241                         if (graph->chunk_commit_data)
 242                                 chunk_repeated = 1;
 243                         else
 244                                 graph->chunk_commit_data = data + chunk_offset;
 245                         break;
 246
 247                 case GRAPH_CHUNKID_EXTRAEDGES:
 248                         if (graph->chunk_extra_edges)
 249                                 chunk_repeated = 1;
 250                         else
 251                                 graph->chunk_extra_edges = data + chunk_offset;
 252                         break;
 253                 }
 254
 255                 if (chunk_repeated) {
 256                         error(_("chunk id %08x appears multiple times"), chunk_id);
 257                         free(graph);
 258                         return NULL;
 259                 }
 260
 261                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 262                 {
 263                         graph->num_commits = (chunk_offset - last_chunk_offset)
 264                                              / graph->hash_len;
 265                 }
 266
 267                 last_chunk_id = chunk_id;
 268                 last_chunk_offset = chunk_offset;
 269         }
 270
 271         if (verify_commit_graph_lite(graph))
 272                 return NULL;
 273
 274         return graph;
 275 }
 276
 277 static struct commit_graph *load_commit_graph_one(const char *graph_file)
 278 {
 279
 280         struct stat st;
 281         int fd;
 282         int open_ok = open_commit_graph(graph_file, &fd, &st);
 283
 284         if (!open_ok)
 285                 return NULL;
 286
 287         return load_commit_graph_one_fd_st(graph_file, fd, &st);
 288 }
 289
 290 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 291 {
 292         char *graph_name;
 293
 294         if (r->objects->commit_graph)
 295                 return;
 296
 297         graph_name = get_commit_graph_filename(obj_dir);
 298         r->objects->commit_graph =
 299                 load_commit_graph_one(graph_name);
 300
 301         FREE_AND_NULL(graph_name);
 302 }
 303
 304 /*
 305  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 306  *
 307  * On the first invocation, this function attemps to load the commit
 308  * graph if the_repository is configured to have one.
 309  */
 310 static int prepare_commit_graph(struct repository *r)
 311 {
 312         struct object_directory *odb;
 313         int config_value;
 314
 315         if (r->objects->commit_graph_attempted)
 316                 return !!r->objects->commit_graph;
 317         r->objects->commit_graph_attempted = 1;
 318
 319         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 320             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 321             !config_value))
 322                 /*
 323                  * This repository is not configured to use commit graphs, so
 324                  * do not load one. (But report commit_graph_attempted anyway
 325                  * so that commit graph loading is not attempted again for this
 326                  * repository.)
 327                  */
 328                 return 0;
 329
 330         if (!commit_graph_compatible(r))
 331                 return 0;
 332
 333         prepare_alt_odb(r);
 334         for (odb = r->objects->odb;
 335              !r->objects->commit_graph && odb;
 336              odb = odb->next)
 337                 prepare_commit_graph_one(r, odb->path);
 338         return !!r->objects->commit_graph;
 339 }
 340
 341 int generation_numbers_enabled(struct repository *r)
 342 {
 343         uint32_t first_generation;
 344         struct commit_graph *g;
 345         if (!prepare_commit_graph(r))
 346                return 0;
 347
 348         g = r->objects->commit_graph;
 349
 350         if (!g->num_commits)
 351                 return 0;
 352
 353         first_generation = get_be32(g->chunk_commit_data +
 354                                     g->hash_len + 8) >> 2;
 355
 356         return !!first_generation;
 357 }
 358
 359 void close_commit_graph(struct repository *r)
 360 {
 361         free_commit_graph(r->objects->commit_graph);
 362         r->objects->commit_graph = NULL;
 363 }
 364
 365 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 366 {
 367         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 368                             g->chunk_oid_lookup, g->hash_len, pos);
 369 }
 370
 371 static struct commit_list **insert_parent_or_die(struct repository *r,
 372                                                  struct commit_graph *g,
 373                                                  uint64_t pos,
 374                                                  struct commit_list **pptr)
 375 {
 376         struct commit *c;
 377         struct object_id oid;
 378
 379         if (pos >= g->num_commits)
 380                 die("invalid parent position %"PRIu64, pos);
 381
 382         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 383         c = lookup_commit(r, &oid);
 384         if (!c)
 385                 die(_("could not find commit %s"), oid_to_hex(&oid));
 386         c->graph_pos = pos;
 387         return &commit_list_insert(c, pptr)->next;
 388 }
 389
 390 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 391 {
 392         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 393         item->graph_pos = pos;
 394         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 395 }
 396
 397 static int fill_commit_in_graph(struct repository *r,
 398                                 struct commit *item,
 399                                 struct commit_graph *g, uint32_t pos)
 400 {
 401         uint32_t edge_value;
 402         uint32_t *parent_data_ptr;
 403         uint64_t date_low, date_high;
 404         struct commit_list **pptr;
 405         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 406
 407         item->object.parsed = 1;
 408         item->graph_pos = pos;
 409
 410         item->maybe_tree = NULL;
 411
 412         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 413         date_low = get_be32(commit_data + g->hash_len + 12);
 414         item->date = (timestamp_t)((date_high << 32) | date_low);
 415
 416         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 417
 418         pptr = &item->parents;
 419
 420         edge_value = get_be32(commit_data + g->hash_len);
 421         if (edge_value == GRAPH_PARENT_NONE)
 422                 return 1;
 423         pptr = insert_parent_or_die(r, g, edge_value, pptr);
 424
 425         edge_value = get_be32(commit_data + g->hash_len + 4);
 426         if (edge_value == GRAPH_PARENT_NONE)
 427                 return 1;
 428         if (!(edge_value & GRAPH_EXTRA_EDGES_NEEDED)) {
 429                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
 430                 return 1;
 431         }
 432
 433         parent_data_ptr = (uint32_t*)(g->chunk_extra_edges +
 434                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 435         do {
 436                 edge_value = get_be32(parent_data_ptr);
 437                 pptr = insert_parent_or_die(r, g,
 438                                             edge_value & GRAPH_EDGE_LAST_MASK,
 439                                             pptr);
 440                 parent_data_ptr++;
 441         } while (!(edge_value & GRAPH_LAST_EDGE));
 442
 443         return 1;
 444 }
 445
 446 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 447 {
 448         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 449                 *pos = item->graph_pos;
 450                 return 1;
 451         } else {
 452                 return bsearch_graph(g, &(item->object.oid), pos);
 453         }
 454 }
 455
 456 static int parse_commit_in_graph_one(struct repository *r,
 457                                      struct commit_graph *g,
 458                                      struct commit *item)
 459 {
 460         uint32_t pos;
 461
 462         if (item->object.parsed)
 463                 return 1;
 464
 465         if (find_commit_in_graph(item, g, &pos))
 466                 return fill_commit_in_graph(r, item, g, pos);
 467
 468         return 0;
 469 }
 470
 471 int parse_commit_in_graph(struct repository *r, struct commit *item)
 472 {
 473         if (!prepare_commit_graph(r))
 474                 return 0;
 475         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
 476 }
 477
 478 void load_commit_graph_info(struct repository *r, struct commit *item)
 479 {
 480         uint32_t pos;
 481         if (!prepare_commit_graph(r))
 482                 return;
 483         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 484                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 485 }
 486
 487 static struct tree *load_tree_for_commit(struct repository *r,
 488                                          struct commit_graph *g,
 489                                          struct commit *c)
 490 {
 491         struct object_id oid;
 492         const unsigned char *commit_data = g->chunk_commit_data +
 493                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 494
 495         hashcpy(oid.hash, commit_data);
 496         c->maybe_tree = lookup_tree(r, &oid);
 497
 498         return c->maybe_tree;
 499 }
 500
 501 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
 502                                                  struct commit_graph *g,
 503                                                  const struct commit *c)
 504 {
 505         if (c->maybe_tree)
 506                 return c->maybe_tree;
 507         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 508                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 509
 510         return load_tree_for_commit(r, g, (struct commit *)c);
 511 }
 512
 513 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 514 {
 515         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
 516 }
 517
 518 static void write_graph_chunk_fanout(struct hashfile *f,
 519                                      struct commit **commits,
 520                                      int nr_commits,
 521                                      struct progress *progress,
 522                                      uint64_t *progress_cnt)
 523 {
 524         int i, count = 0;
 525         struct commit **list = commits;
 526
 527         /*
 528          * Write the first-level table (the list is sorted,
 529          * but we use a 256-entry lookup to be able to avoid
 530          * having to do eight extra binary search iterations).
 531          */
 532         for (i = 0; i < 256; i++) {
 533                 while (count < nr_commits) {
 534                         if ((*list)->object.oid.hash[0] != i)
 535                                 break;
 536                         display_progress(progress, ++*progress_cnt);
 537                         count++;
 538                         list++;
 539                 }
 540
 541                 hashwrite_be32(f, count);
 542         }
 543 }
 544
 545 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 546                                    struct commit **commits, int nr_commits,
 547                                    struct progress *progress,
 548                                    uint64_t *progress_cnt)
 549 {
 550         struct commit **list = commits;
 551         int count;
 552         for (count = 0; count < nr_commits; count++, list++) {
 553                 display_progress(progress, ++*progress_cnt);
 554                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 555         }
 556 }
 557
 558 static const unsigned char *commit_to_sha1(size_t index, void *table)
 559 {
 560         struct commit **commits = table;
 561         return commits[index]->object.oid.hash;
 562 }
 563
 564 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 565                                    struct commit **commits, int nr_commits,
 566                                    struct progress *progress,
 567                                    uint64_t *progress_cnt)
 568 {
 569         struct commit **list = commits;
 570         struct commit **last = commits + nr_commits;
 571         uint32_t num_extra_edges = 0;
 572
 573         while (list < last) {
 574                 struct commit_list *parent;
 575                 int edge_value;
 576                 uint32_t packedDate[2];
 577                 display_progress(progress, ++*progress_cnt);
 578
 579                 parse_commit(*list);
 580                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 581
 582                 parent = (*list)->parents;
 583
 584                 if (!parent)
 585                         edge_value = GRAPH_PARENT_NONE;
 586                 else {
 587                         edge_value = sha1_pos(parent->item->object.oid.hash,
 588                                               commits,
 589                                               nr_commits,
 590                                               commit_to_sha1);
 591
 592                         if (edge_value < 0)
 593                                 BUG("missing parent %s for commit %s",
 594                                     oid_to_hex(&parent->item->object.oid),
 595                                     oid_to_hex(&(*list)->object.oid));
 596                 }
 597
 598                 hashwrite_be32(f, edge_value);
 599
 600                 if (parent)
 601                         parent = parent->next;
 602
 603                 if (!parent)
 604                         edge_value = GRAPH_PARENT_NONE;
 605                 else if (parent->next)
 606                         edge_value = GRAPH_EXTRA_EDGES_NEEDED | num_extra_edges;
 607                 else {
 608                         edge_value = sha1_pos(parent->item->object.oid.hash,
 609                                               commits,
 610                                               nr_commits,
 611                                               commit_to_sha1);
 612                         if (edge_value < 0)
 613                                 BUG("missing parent %s for commit %s",
 614                                     oid_to_hex(&parent->item->object.oid),
 615                                     oid_to_hex(&(*list)->object.oid));
 616                 }
 617
 618                 hashwrite_be32(f, edge_value);
 619
 620                 if (edge_value & GRAPH_EXTRA_EDGES_NEEDED) {
 621                         do {
 622                                 num_extra_edges++;
 623                                 parent = parent->next;
 624                         } while (parent);
 625                 }
 626
 627                 if (sizeof((*list)->date) > 4)
 628                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 629                 else
 630                         packedDate[0] = 0;
 631
 632                 packedDate[0] |= htonl((*list)->generation << 2);
 633
 634                 packedDate[1] = htonl((*list)->date);
 635                 hashwrite(f, packedDate, 8);
 636
 637                 list++;
 638         }
 639 }
 640
 641 static void write_graph_chunk_extra_edges(struct hashfile *f,
 642                                           struct commit **commits,
 643                                           int nr_commits,
 644                                           struct progress *progress,
 645                                           uint64_t *progress_cnt)
 646 {
 647         struct commit **list = commits;
 648         struct commit **last = commits + nr_commits;
 649         struct commit_list *parent;
 650
 651         while (list < last) {
 652                 int num_parents = 0;
 653
 654                 display_progress(progress, ++*progress_cnt);
 655
 656                 for (parent = (*list)->parents; num_parents < 3 && parent;
 657                      parent = parent->next)
 658                         num_parents++;
 659
 660                 if (num_parents <= 2) {
 661                         list++;
 662                         continue;
 663                 }
 664
 665                 /* Since num_parents > 2, this initializer is safe. */
 666                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 667                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 668                                                   commits,
 669                                                   nr_commits,
 670                                                   commit_to_sha1);
 671
 672                         if (edge_value < 0)
 673                                 BUG("missing parent %s for commit %s",
 674                                     oid_to_hex(&parent->item->object.oid),
 675                                     oid_to_hex(&(*list)->object.oid));
 676                         else if (!parent->next)
 677                                 edge_value |= GRAPH_LAST_EDGE;
 678
 679                         hashwrite_be32(f, edge_value);
 680                 }
 681
 682                 list++;
 683         }
 684 }
 685
 686 static int commit_compare(const void *_a, const void *_b)
 687 {
 688         const struct object_id *a = (const struct object_id *)_a;
 689         const struct object_id *b = (const struct object_id *)_b;
 690         return oidcmp(a, b);
 691 }
 692
 693 struct packed_commit_list {
 694         struct commit **list;
 695         int nr;
 696         int alloc;
 697 };
 698
 699 struct packed_oid_list {
 700         struct object_id *list;
 701         int nr;
 702         int alloc;
 703         struct progress *progress;
 704         int progress_done;
 705 };
 706
 707 static int add_packed_commits(const struct object_id *oid,
 708                               struct packed_git *pack,
 709                               uint32_t pos,
 710                               void *data)
 711 {
 712         struct packed_oid_list *list = (struct packed_oid_list*)data;
 713         enum object_type type;
 714         off_t offset = nth_packed_object_offset(pack, pos);
 715         struct object_info oi = OBJECT_INFO_INIT;
 716
 717         if (list->progress)
 718                 display_progress(list->progress, ++list->progress_done);
 719
 720         oi.typep = &type;
 721         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 722                 die(_("unable to get type of object %s"), oid_to_hex(oid));
 723
 724         if (type != OBJ_COMMIT)
 725                 return 0;
 726
 727         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 728         oidcpy(&(list->list[list->nr]), oid);
 729         list->nr++;
 730
 731         return 0;
 732 }
 733
 734 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 735 {
 736         struct commit_list *parent;
 737         for (parent = commit->parents; parent; parent = parent->next) {
 738                 if (!(parent->item->object.flags & UNINTERESTING)) {
 739                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 740                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 741                         oids->nr++;
 742                         parent->item->object.flags |= UNINTERESTING;
 743                 }
 744         }
 745 }
 746
 747 static void close_reachable(struct packed_oid_list *oids, int report_progress)
 748 {
 749         int i;
 750         struct commit *commit;
 751         struct progress *progress = NULL;
 752
 753         if (report_progress)
 754                 progress = start_delayed_progress(
 755                         _("Loading known commits in commit graph"), oids->nr);
 756         for (i = 0; i < oids->nr; i++) {
 757                 display_progress(progress, i + 1);
 758                 commit = lookup_commit(the_repository, &oids->list[i]);
 759                 if (commit)
 760                         commit->object.flags |= UNINTERESTING;
 761         }
 762         stop_progress(&progress);
 763
 764         /*
 765          * As this loop runs, oids->nr may grow, but not more
 766          * than the number of missing commits in the reachable
 767          * closure.
 768          */
 769         if (report_progress)
 770                 progress = start_delayed_progress(
 771                         _("Expanding reachable commits in commit graph"), oids->nr);
 772         for (i = 0; i < oids->nr; i++) {
 773                 display_progress(progress, i + 1);
 774                 commit = lookup_commit(the_repository, &oids->list[i]);
 775
 776                 if (commit && !parse_commit(commit))
 777                         add_missing_parents(oids, commit);
 778         }
 779         stop_progress(&progress);
 780
 781         if (report_progress)
 782                 progress = start_delayed_progress(
 783                         _("Clearing commit marks in commit graph"), oids->nr);
 784         for (i = 0; i < oids->nr; i++) {
 785                 display_progress(progress, i + 1);
 786                 commit = lookup_commit(the_repository, &oids->list[i]);
 787
 788                 if (commit)
 789                         commit->object.flags &= ~UNINTERESTING;
 790         }
 791         stop_progress(&progress);
 792 }
 793
 794 static void compute_generation_numbers(struct packed_commit_list* commits,
 795                                        int report_progress)
 796 {
 797         int i;
 798         struct commit_list *list = NULL;
 799         struct progress *progress = NULL;
 800
 801         if (report_progress)
 802                 progress = start_progress(
 803                         _("Computing commit graph generation numbers"),
 804                         commits->nr);
 805         for (i = 0; i < commits->nr; i++) {
 806                 display_progress(progress, i + 1);
 807                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 808                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 809                         continue;
 810
 811                 commit_list_insert(commits->list[i], &list);
 812                 while (list) {
 813                         struct commit *current = list->item;
 814                         struct commit_list *parent;
 815                         int all_parents_computed = 1;
 816                         uint32_t max_generation = 0;
 817
 818                         for (parent = current->parents; parent; parent = parent->next) {
 819                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 820                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 821                                         all_parents_computed = 0;
 822                                         commit_list_insert(parent->item, &list);
 823                                         break;
 824                                 } else if (parent->item->generation > max_generation) {
 825                                         max_generation = parent->item->generation;
 826                                 }
 827                         }
 828
 829                         if (all_parents_computed) {
 830                                 current->generation = max_generation + 1;
 831                                 pop_commit(&list);
 832
 833                                 if (current->generation > GENERATION_NUMBER_MAX)
 834                                         current->generation = GENERATION_NUMBER_MAX;
 835                         }
 836                 }
 837         }
 838         stop_progress(&progress);
 839 }
 840
 841 static int add_ref_to_list(const char *refname,
 842                            const struct object_id *oid,
 843                            int flags, void *cb_data)
 844 {
 845         struct string_list *list = (struct string_list *)cb_data;
 846
 847         string_list_append(list, oid_to_hex(oid));
 848         return 0;
 849 }
 850
 851 void write_commit_graph_reachable(const char *obj_dir, int append,
 852                                   int report_progress)
 853 {
 854         struct string_list list = STRING_LIST_INIT_DUP;
 855
 856         for_each_ref(add_ref_to_list, &list);
 857         write_commit_graph(obj_dir, NULL, &list, append, report_progress);
 858
 859         string_list_clear(&list, 0);
 860 }
 861
 862 void write_commit_graph(const char *obj_dir,
 863                         struct string_list *pack_indexes,
 864                         struct string_list *commit_hex,
 865                         int append, int report_progress)
 866 {
 867         struct packed_oid_list oids;
 868         struct packed_commit_list commits;
 869         struct hashfile *f;
 870         uint32_t i, count_distinct = 0;
 871         char *graph_name;
 872         struct lock_file lk = LOCK_INIT;
 873         uint32_t chunk_ids[5];
 874         uint64_t chunk_offsets[5];
 875         int num_chunks;
 876         int num_extra_edges;
 877         struct commit_list *parent;
 878         struct progress *progress = NULL;
 879         const unsigned hashsz = the_hash_algo->rawsz;
 880         uint64_t progress_cnt = 0;
 881         struct strbuf progress_title = STRBUF_INIT;
 882         unsigned long approx_nr_objects;
 883
 884         if (!commit_graph_compatible(the_repository))
 885                 return;
 886
 887         oids.nr = 0;
 888         approx_nr_objects = approximate_object_count();
 889         oids.alloc = approx_nr_objects / 32;
 890         oids.progress = NULL;
 891         oids.progress_done = 0;
 892
 893         if (append) {
 894                 prepare_commit_graph_one(the_repository, obj_dir);
 895                 if (the_repository->objects->commit_graph)
 896                         oids.alloc += the_repository->objects->commit_graph->num_commits;
 897         }
 898
 899         if (oids.alloc < 1024)
 900                 oids.alloc = 1024;
 901         ALLOC_ARRAY(oids.list, oids.alloc);
 902
 903         if (append && the_repository->objects->commit_graph) {
 904                 struct commit_graph *commit_graph =
 905                         the_repository->objects->commit_graph;
 906                 for (i = 0; i < commit_graph->num_commits; i++) {
 907                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 908                                 commit_graph->hash_len * i;
 909                         hashcpy(oids.list[oids.nr++].hash, hash);
 910                 }
 911         }
 912
 913         if (pack_indexes) {
 914                 struct strbuf packname = STRBUF_INIT;
 915                 int dirlen;
 916                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 917                 dirlen = packname.len;
 918                 if (report_progress) {
 919                         strbuf_addf(&progress_title,
 920                                     Q_("Finding commits for commit graph in %d pack",
 921                                        "Finding commits for commit graph in %d packs",
 922                                        pack_indexes->nr),
 923                                     pack_indexes->nr);
 924                         oids.progress = start_delayed_progress(progress_title.buf, 0);
 925                         oids.progress_done = 0;
 926                 }
 927                 for (i = 0; i < pack_indexes->nr; i++) {
 928                         struct packed_git *p;
 929                         strbuf_setlen(&packname, dirlen);
 930                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 931                         p = add_packed_git(packname.buf, packname.len, 1);
 932                         if (!p)
 933                                 die(_("error adding pack %s"), packname.buf);
 934                         if (open_pack_index(p))
 935                                 die(_("error opening index for %s"), packname.buf);
 936                         for_each_object_in_pack(p, add_packed_commits, &oids,
 937                                                 FOR_EACH_OBJECT_PACK_ORDER);
 938                         close_pack(p);
 939                         free(p);
 940                 }
 941                 stop_progress(&oids.progress);
 942                 strbuf_reset(&progress_title);
 943                 strbuf_release(&packname);
 944         }
 945
 946         if (commit_hex) {
 947                 if (report_progress) {
 948                         strbuf_addf(&progress_title,
 949                                     Q_("Finding commits for commit graph from %d ref",
 950                                        "Finding commits for commit graph from %d refs",
 951                                        commit_hex->nr),
 952                                     commit_hex->nr);
 953                         progress = start_delayed_progress(progress_title.buf,
 954                                                           commit_hex->nr);
 955                 }
 956                 for (i = 0; i < commit_hex->nr; i++) {
 957                         const char *end;
 958                         struct object_id oid;
 959                         struct commit *result;
 960
 961                         display_progress(progress, i + 1);
 962                         if (commit_hex->items[i].string &&
 963                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 964                                 continue;
 965
 966                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
 967
 968                         if (result) {
 969                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 970                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 971                                 oids.nr++;
 972                         }
 973                 }
 974                 stop_progress(&progress);
 975                 strbuf_reset(&progress_title);
 976         }
 977
 978         if (!pack_indexes && !commit_hex) {
 979                 if (report_progress)
 980                         oids.progress = start_delayed_progress(
 981                                 _("Finding commits for commit graph among packed objects"),
 982                                 approx_nr_objects);
 983                 for_each_packed_object(add_packed_commits, &oids,
 984                                        FOR_EACH_OBJECT_PACK_ORDER);
 985                 if (oids.progress_done < approx_nr_objects)
 986                         display_progress(oids.progress, approx_nr_objects);
 987                 stop_progress(&oids.progress);
 988         }
 989
 990         close_reachable(&oids, report_progress);
 991
 992         if (report_progress)
 993                 progress = start_delayed_progress(
 994                         _("Counting distinct commits in commit graph"),
 995                         oids.nr);
 996         display_progress(progress, 0); /* TODO: Measure QSORT() progress */
 997         QSORT(oids.list, oids.nr, commit_compare);
 998         count_distinct = 1;
 999         for (i = 1; i < oids.nr; i++) {
1000                 display_progress(progress, i + 1);
1001                 if (!oideq(&oids.list[i - 1], &oids.list[i]))
1002                         count_distinct++;
1003         }
1004         stop_progress(&progress);
1005
1006         if (count_distinct >= GRAPH_EDGE_LAST_MASK)
1007                 die(_("the commit graph format cannot write %d commits"), count_distinct);
1008
1009         commits.nr = 0;
1010         commits.alloc = count_distinct;
1011         ALLOC_ARRAY(commits.list, commits.alloc);
1012
1013         num_extra_edges = 0;
1014         if (report_progress)
1015                 progress = start_delayed_progress(
1016                         _("Finding extra edges in commit graph"),
1017                         oids.nr);
1018         for (i = 0; i < oids.nr; i++) {
1019                 int num_parents = 0;
1020                 display_progress(progress, i + 1);
1021                 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
1022                         continue;
1023
1024                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
1025                 parse_commit(commits.list[commits.nr]);
1026
1027                 for (parent = commits.list[commits.nr]->parents;
1028                      parent; parent = parent->next)
1029                         num_parents++;
1030
1031                 if (num_parents > 2)
1032                         num_extra_edges += num_parents - 1;
1033
1034                 commits.nr++;
1035         }
1036         num_chunks = num_extra_edges ? 4 : 3;
1037         stop_progress(&progress);
1038
1039         if (commits.nr >= GRAPH_EDGE_LAST_MASK)
1040                 die(_("too many commits to write graph"));
1041
1042         compute_generation_numbers(&commits, report_progress);
1043
1044         graph_name = get_commit_graph_filename(obj_dir);
1045         if (safe_create_leading_directories(graph_name)) {
1046                 UNLEAK(graph_name);
1047                 die_errno(_("unable to create leading directories of %s"),
1048                           graph_name);
1049         }
1050
1051         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
1052         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
1053
1054         hashwrite_be32(f, GRAPH_SIGNATURE);
1055
1056         hashwrite_u8(f, GRAPH_VERSION);
1057         hashwrite_u8(f, oid_version());
1058         hashwrite_u8(f, num_chunks);
1059         hashwrite_u8(f, 0); /* unused padding byte */
1060
1061         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
1062         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
1063         chunk_ids[2] = GRAPH_CHUNKID_DATA;
1064         if (num_extra_edges)
1065                 chunk_ids[3] = GRAPH_CHUNKID_EXTRAEDGES;
1066         else
1067                 chunk_ids[3] = 0;
1068         chunk_ids[4] = 0;
1069
1070         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
1071         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
1072         chunk_offsets[2] = chunk_offsets[1] + hashsz * commits.nr;
1073         chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * commits.nr;
1074         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
1075
1076         for (i = 0; i <= num_chunks; i++) {
1077                 uint32_t chunk_write[3];
1078
1079                 chunk_write[0] = htonl(chunk_ids[i]);
1080                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
1081                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
1082                 hashwrite(f, chunk_write, 12);
1083         }
1084
1085         if (report_progress) {
1086                 strbuf_addf(&progress_title,
1087                             Q_("Writing out commit graph in %d pass",
1088                                "Writing out commit graph in %d passes",
1089                                num_chunks),
1090                             num_chunks);
1091                 progress = start_delayed_progress(
1092                         progress_title.buf,
1093                         num_chunks * commits.nr);
1094         }
1095         write_graph_chunk_fanout(f, commits.list, commits.nr, progress, &progress_cnt);
1096         write_graph_chunk_oids(f, hashsz, commits.list, commits.nr, progress, &progress_cnt);
1097         write_graph_chunk_data(f, hashsz, commits.list, commits.nr, progress, &progress_cnt);
1098         if (num_extra_edges)
1099                 write_graph_chunk_extra_edges(f, commits.list, commits.nr, progress, &progress_cnt);
1100         stop_progress(&progress);
1101         strbuf_release(&progress_title);
1102
1103         close_commit_graph(the_repository);
1104         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
1105         commit_lock_file(&lk);
1106
1107         free(graph_name);
1108         free(commits.list);
1109         free(oids.list);
1110 }
1111
1112 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
1113 static int verify_commit_graph_error;
1114
1115 static void graph_report(const char *fmt, ...)
1116 {
1117         va_list ap;
1118
1119         verify_commit_graph_error = 1;
1120         va_start(ap, fmt);
1121         vfprintf(stderr, fmt, ap);
1122         fprintf(stderr, "\n");
1123         va_end(ap);
1124 }
1125
1126 #define GENERATION_ZERO_EXISTS 1
1127 #define GENERATION_NUMBER_EXISTS 2
1128
1129 int verify_commit_graph(struct repository *r, struct commit_graph *g)
1130 {
1131         uint32_t i, cur_fanout_pos = 0;
1132         struct object_id prev_oid, cur_oid, checksum;
1133         int generation_zero = 0;
1134         struct hashfile *f;
1135         int devnull;
1136         struct progress *progress = NULL;
1137
1138         if (!g) {
1139                 graph_report("no commit-graph file loaded");
1140                 return 1;
1141         }
1142
1143         verify_commit_graph_error = verify_commit_graph_lite(g);
1144         if (verify_commit_graph_error)
1145                 return verify_commit_graph_error;
1146
1147         devnull = open("/dev/null", O_WRONLY);
1148         f = hashfd(devnull, NULL);
1149         hashwrite(f, g->data, g->data_len - g->hash_len);
1150         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1151         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1152                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1153                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1154         }
1155
1156         for (i = 0; i < g->num_commits; i++) {
1157                 struct commit *graph_commit;
1158
1159                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1160
1161                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1162                         graph_report("commit-graph has incorrect OID order: %s then %s",
1163                                      oid_to_hex(&prev_oid),
1164                                      oid_to_hex(&cur_oid));
1165
1166                 oidcpy(&prev_oid, &cur_oid);
1167
1168                 while (cur_oid.hash[0] > cur_fanout_pos) {
1169                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1170
1171                         if (i != fanout_value)
1172                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1173                                              cur_fanout_pos, fanout_value, i);
1174                         cur_fanout_pos++;
1175                 }
1176
1177                 graph_commit = lookup_commit(r, &cur_oid);
1178                 if (!parse_commit_in_graph_one(r, g, graph_commit))
1179                         graph_report("failed to parse %s from commit-graph",
1180                                      oid_to_hex(&cur_oid));
1181         }
1182
1183         while (cur_fanout_pos < 256) {
1184                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1185
1186                 if (g->num_commits != fanout_value)
1187                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1188                                      cur_fanout_pos, fanout_value, i);
1189
1190                 cur_fanout_pos++;
1191         }
1192
1193         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1194                 return verify_commit_graph_error;
1195
1196         progress = start_progress(_("Verifying commits in commit graph"),
1197                                   g->num_commits);
1198         for (i = 0; i < g->num_commits; i++) {
1199                 struct commit *graph_commit, *odb_commit;
1200                 struct commit_list *graph_parents, *odb_parents;
1201                 uint32_t max_generation = 0;
1202
1203                 display_progress(progress, i + 1);
1204                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1205
1206                 graph_commit = lookup_commit(r, &cur_oid);
1207                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1208                 if (parse_commit_internal(odb_commit, 0, 0)) {
1209                         graph_report("failed to parse %s from object database",
1210                                      oid_to_hex(&cur_oid));
1211                         continue;
1212                 }
1213
1214                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1215                            get_commit_tree_oid(odb_commit)))
1216                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1217                                      oid_to_hex(&cur_oid),
1218                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1219                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1220
1221                 graph_parents = graph_commit->parents;
1222                 odb_parents = odb_commit->parents;
1223
1224                 while (graph_parents) {
1225                         if (odb_parents == NULL) {
1226                                 graph_report("commit-graph parent list for commit %s is too long",
1227                                              oid_to_hex(&cur_oid));
1228                                 break;
1229                         }
1230
1231                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1232                                 graph_report("commit-graph parent for %s is %s != %s",
1233                                              oid_to_hex(&cur_oid),
1234                                              oid_to_hex(&graph_parents->item->object.oid),
1235                                              oid_to_hex(&odb_parents->item->object.oid));
1236
1237                         if (graph_parents->item->generation > max_generation)
1238                                 max_generation = graph_parents->item->generation;
1239
1240                         graph_parents = graph_parents->next;
1241                         odb_parents = odb_parents->next;
1242                 }
1243
1244                 if (odb_parents != NULL)
1245                         graph_report("commit-graph parent list for commit %s terminates early",
1246                                      oid_to_hex(&cur_oid));
1247
1248                 if (!graph_commit->generation) {
1249                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1250                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1251                                              oid_to_hex(&cur_oid));
1252                         generation_zero = GENERATION_ZERO_EXISTS;
1253                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1254                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1255                                      oid_to_hex(&cur_oid));
1256
1257                 if (generation_zero == GENERATION_ZERO_EXISTS)
1258                         continue;
1259
1260                 /*
1261                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1262                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1263                  * extra logic in the following condition.
1264                  */
1265                 if (max_generation == GENERATION_NUMBER_MAX)
1266                         max_generation--;
1267
1268                 if (graph_commit->generation != max_generation + 1)
1269                         graph_report("commit-graph generation for commit %s is %u != %u",
1270                                      oid_to_hex(&cur_oid),
1271                                      graph_commit->generation,
1272                                      max_generation + 1);
1273
1274                 if (graph_commit->date != odb_commit->date)
1275                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1276                                      oid_to_hex(&cur_oid),
1277                                      graph_commit->date,
1278                                      odb_commit->date);
1279         }
1280         stop_progress(&progress);
1281
1282         return verify_commit_graph_error;
1283 }
1284
1285 void free_commit_graph(struct commit_graph *g)
1286 {
1287         if (!g)
1288                 return;
1289         if (g->graph_fd >= 0) {
1290                 munmap((void *)g->data, g->data_len);
1291                 g->data = NULL;
1292                 close(g->graph_fd);
1293         }
1294         free(g);
1295 }