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