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

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "dir.h"
   4 #include "git-compat-util.h"
   5 #include "lockfile.h"
   6 #include "pack.h"
   7 #include "packfile.h"
   8 #include "commit.h"
   9 #include "object.h"
  10 #include "refs.h"
  11 #include "revision.h"
  12 #include "sha1-lookup.h"
  13 #include "commit-graph.h"
  14 #include "object-store.h"
  15 #include "alloc.h"
  16 #include "hashmap.h"
  17 #include "replace-object.h"
  18 #include "progress.h"
  19
  20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  24 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  25
  26 #define GRAPH_DATA_WIDTH (the_hash_algo->rawsz + 16)
  27
  28 #define GRAPH_VERSION_1 0x1
  29 #define GRAPH_VERSION GRAPH_VERSION_1
  30
  31 #define GRAPH_OCTOPUS_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 struct commit_graph *load_commit_graph_one(const char *graph_file)
  84 {
  85         void *graph_map;
  86         const unsigned char *data, *chunk_lookup;
  87         size_t graph_size;
  88         struct stat st;
  89         uint32_t i;
  90         struct commit_graph *graph;
  91         int fd = git_open(graph_file);
  92         uint64_t last_chunk_offset;
  93         uint32_t last_chunk_id;
  94         uint32_t graph_signature;
  95         unsigned char graph_version, hash_version;
  96
  97         if (fd < 0)
  98                 return NULL;
  99         if (fstat(fd, &st)) {
 100                 close(fd);
 101                 return NULL;
 102         }
 103         graph_size = xsize_t(st.st_size);
 104
 105         if (graph_size < GRAPH_MIN_SIZE) {
 106                 close(fd);
 107                 die(_("graph file %s is too small"), graph_file);
 108         }
 109         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
 110         data = (const unsigned char *)graph_map;
 111
 112         graph_signature = get_be32(data);
 113         if (graph_signature != GRAPH_SIGNATURE) {
 114                 error(_("graph signature %X does not match signature %X"),
 115                       graph_signature, GRAPH_SIGNATURE);
 116                 goto cleanup_fail;
 117         }
 118
 119         graph_version = *(unsigned char*)(data + 4);
 120         if (graph_version != GRAPH_VERSION) {
 121                 error(_("graph version %X does not match version %X"),
 122                       graph_version, GRAPH_VERSION);
 123                 goto cleanup_fail;
 124         }
 125
 126         hash_version = *(unsigned char*)(data + 5);
 127         if (hash_version != oid_version()) {
 128                 error(_("hash version %X does not match version %X"),
 129                       hash_version, oid_version());
 130                 goto cleanup_fail;
 131         }
 132
 133         graph = alloc_commit_graph();
 134
 135         graph->hash_len = the_hash_algo->rawsz;
 136         graph->num_chunks = *(unsigned char*)(data + 6);
 137         graph->graph_fd = fd;
 138         graph->data = graph_map;
 139         graph->data_len = graph_size;
 140
 141         last_chunk_id = 0;
 142         last_chunk_offset = 8;
 143         chunk_lookup = data + 8;
 144         for (i = 0; i < graph->num_chunks; i++) {
 145                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 146                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 147                 int chunk_repeated = 0;
 148
 149                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 150
 151                 if (chunk_offset > graph_size - the_hash_algo->rawsz) {
 152                         error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 153                               (uint32_t)chunk_offset);
 154                         goto cleanup_fail;
 155                 }
 156
 157                 switch (chunk_id) {
 158                 case GRAPH_CHUNKID_OIDFANOUT:
 159                         if (graph->chunk_oid_fanout)
 160                                 chunk_repeated = 1;
 161                         else
 162                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 163                         break;
 164
 165                 case GRAPH_CHUNKID_OIDLOOKUP:
 166                         if (graph->chunk_oid_lookup)
 167                                 chunk_repeated = 1;
 168                         else
 169                                 graph->chunk_oid_lookup = data + chunk_offset;
 170                         break;
 171
 172                 case GRAPH_CHUNKID_DATA:
 173                         if (graph->chunk_commit_data)
 174                                 chunk_repeated = 1;
 175                         else
 176                                 graph->chunk_commit_data = data + chunk_offset;
 177                         break;
 178
 179                 case GRAPH_CHUNKID_LARGEEDGES:
 180                         if (graph->chunk_large_edges)
 181                                 chunk_repeated = 1;
 182                         else
 183                                 graph->chunk_large_edges = data + chunk_offset;
 184                         break;
 185                 }
 186
 187                 if (chunk_repeated) {
 188                         error(_("chunk id %08x appears multiple times"), chunk_id);
 189                         goto cleanup_fail;
 190                 }
 191
 192                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 193                 {
 194                         graph->num_commits = (chunk_offset - last_chunk_offset)
 195                                              / graph->hash_len;
 196                 }
 197
 198                 last_chunk_id = chunk_id;
 199                 last_chunk_offset = chunk_offset;
 200         }
 201
 202         return graph;
 203
 204 cleanup_fail:
 205         munmap(graph_map, graph_size);
 206         close(fd);
 207         exit(1);
 208 }
 209
 210 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 211 {
 212         char *graph_name;
 213
 214         if (r->objects->commit_graph)
 215                 return;
 216
 217         graph_name = get_commit_graph_filename(obj_dir);
 218         r->objects->commit_graph =
 219                 load_commit_graph_one(graph_name);
 220
 221         FREE_AND_NULL(graph_name);
 222 }
 223
 224 /*
 225  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 226  *
 227  * On the first invocation, this function attemps to load the commit
 228  * graph if the_repository is configured to have one.
 229  */
 230 static int prepare_commit_graph(struct repository *r)
 231 {
 232         struct object_directory *odb;
 233         int config_value;
 234
 235         if (r->objects->commit_graph_attempted)
 236                 return !!r->objects->commit_graph;
 237         r->objects->commit_graph_attempted = 1;
 238
 239         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 240             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 241             !config_value))
 242                 /*
 243                  * This repository is not configured to use commit graphs, so
 244                  * do not load one. (But report commit_graph_attempted anyway
 245                  * so that commit graph loading is not attempted again for this
 246                  * repository.)
 247                  */
 248                 return 0;
 249
 250         if (!commit_graph_compatible(r))
 251                 return 0;
 252
 253         prepare_alt_odb(r);
 254         for (odb = r->objects->odb;
 255              !r->objects->commit_graph && odb;
 256              odb = odb->next)
 257                 prepare_commit_graph_one(r, odb->path);
 258         return !!r->objects->commit_graph;
 259 }
 260
 261 int generation_numbers_enabled(struct repository *r)
 262 {
 263         uint32_t first_generation;
 264         struct commit_graph *g;
 265         if (!prepare_commit_graph(r))
 266                return 0;
 267
 268         g = r->objects->commit_graph;
 269
 270         if (!g->num_commits)
 271                 return 0;
 272
 273         first_generation = get_be32(g->chunk_commit_data +
 274                                     g->hash_len + 8) >> 2;
 275
 276         return !!first_generation;
 277 }
 278
 279 void close_commit_graph(struct repository *r)
 280 {
 281         free_commit_graph(r->objects->commit_graph);
 282         r->objects->commit_graph = NULL;
 283 }
 284
 285 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 286 {
 287         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 288                             g->chunk_oid_lookup, g->hash_len, pos);
 289 }
 290
 291 static struct commit_list **insert_parent_or_die(struct repository *r,
 292                                                  struct commit_graph *g,
 293                                                  uint64_t pos,
 294                                                  struct commit_list **pptr)
 295 {
 296         struct commit *c;
 297         struct object_id oid;
 298
 299         if (pos >= g->num_commits)
 300                 die("invalid parent position %"PRIu64, pos);
 301
 302         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 303         c = lookup_commit(r, &oid);
 304         if (!c)
 305                 die(_("could not find commit %s"), oid_to_hex(&oid));
 306         c->graph_pos = pos;
 307         return &commit_list_insert(c, pptr)->next;
 308 }
 309
 310 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 311 {
 312         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 313         item->graph_pos = pos;
 314         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 315 }
 316
 317 static int fill_commit_in_graph(struct repository *r,
 318                                 struct commit *item,
 319                                 struct commit_graph *g, uint32_t pos)
 320 {
 321         uint32_t edge_value;
 322         uint32_t *parent_data_ptr;
 323         uint64_t date_low, date_high;
 324         struct commit_list **pptr;
 325         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 326
 327         item->object.parsed = 1;
 328         item->graph_pos = pos;
 329
 330         item->maybe_tree = NULL;
 331
 332         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 333         date_low = get_be32(commit_data + g->hash_len + 12);
 334         item->date = (timestamp_t)((date_high << 32) | date_low);
 335
 336         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 337
 338         pptr = &item->parents;
 339
 340         edge_value = get_be32(commit_data + g->hash_len);
 341         if (edge_value == GRAPH_PARENT_NONE)
 342                 return 1;
 343         pptr = insert_parent_or_die(r, g, edge_value, pptr);
 344
 345         edge_value = get_be32(commit_data + g->hash_len + 4);
 346         if (edge_value == GRAPH_PARENT_NONE)
 347                 return 1;
 348         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 349                 pptr = insert_parent_or_die(r, g, edge_value, pptr);
 350                 return 1;
 351         }
 352
 353         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 354                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 355         do {
 356                 edge_value = get_be32(parent_data_ptr);
 357                 pptr = insert_parent_or_die(r, g,
 358                                             edge_value & GRAPH_EDGE_LAST_MASK,
 359                                             pptr);
 360                 parent_data_ptr++;
 361         } while (!(edge_value & GRAPH_LAST_EDGE));
 362
 363         return 1;
 364 }
 365
 366 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 367 {
 368         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 369                 *pos = item->graph_pos;
 370                 return 1;
 371         } else {
 372                 return bsearch_graph(g, &(item->object.oid), pos);
 373         }
 374 }
 375
 376 static int parse_commit_in_graph_one(struct repository *r,
 377                                      struct commit_graph *g,
 378                                      struct commit *item)
 379 {
 380         uint32_t pos;
 381
 382         if (item->object.parsed)
 383                 return 1;
 384
 385         if (find_commit_in_graph(item, g, &pos))
 386                 return fill_commit_in_graph(r, item, g, pos);
 387
 388         return 0;
 389 }
 390
 391 int parse_commit_in_graph(struct repository *r, struct commit *item)
 392 {
 393         if (!prepare_commit_graph(r))
 394                 return 0;
 395         return parse_commit_in_graph_one(r, r->objects->commit_graph, item);
 396 }
 397
 398 void load_commit_graph_info(struct repository *r, struct commit *item)
 399 {
 400         uint32_t pos;
 401         if (!prepare_commit_graph(r))
 402                 return;
 403         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 404                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 405 }
 406
 407 static struct tree *load_tree_for_commit(struct repository *r,
 408                                          struct commit_graph *g,
 409                                          struct commit *c)
 410 {
 411         struct object_id oid;
 412         const unsigned char *commit_data = g->chunk_commit_data +
 413                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 414
 415         hashcpy(oid.hash, commit_data);
 416         c->maybe_tree = lookup_tree(r, &oid);
 417
 418         return c->maybe_tree;
 419 }
 420
 421 static struct tree *get_commit_tree_in_graph_one(struct repository *r,
 422                                                  struct commit_graph *g,
 423                                                  const struct commit *c)
 424 {
 425         if (c->maybe_tree)
 426                 return c->maybe_tree;
 427         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 428                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 429
 430         return load_tree_for_commit(r, g, (struct commit *)c);
 431 }
 432
 433 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 434 {
 435         return get_commit_tree_in_graph_one(r, r->objects->commit_graph, c);
 436 }
 437
 438 static void write_graph_chunk_fanout(struct hashfile *f,
 439                                      struct commit **commits,
 440                                      int nr_commits)
 441 {
 442         int i, count = 0;
 443         struct commit **list = commits;
 444
 445         /*
 446          * Write the first-level table (the list is sorted,
 447          * but we use a 256-entry lookup to be able to avoid
 448          * having to do eight extra binary search iterations).
 449          */
 450         for (i = 0; i < 256; i++) {
 451                 while (count < nr_commits) {
 452                         if ((*list)->object.oid.hash[0] != i)
 453                                 break;
 454                         count++;
 455                         list++;
 456                 }
 457
 458                 hashwrite_be32(f, count);
 459         }
 460 }
 461
 462 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 463                                    struct commit **commits, int nr_commits)
 464 {
 465         struct commit **list = commits;
 466         int count;
 467         for (count = 0; count < nr_commits; count++, list++)
 468                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 469 }
 470
 471 static const unsigned char *commit_to_sha1(size_t index, void *table)
 472 {
 473         struct commit **commits = table;
 474         return commits[index]->object.oid.hash;
 475 }
 476
 477 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 478                                    struct commit **commits, int nr_commits)
 479 {
 480         struct commit **list = commits;
 481         struct commit **last = commits + nr_commits;
 482         uint32_t num_extra_edges = 0;
 483
 484         while (list < last) {
 485                 struct commit_list *parent;
 486                 int edge_value;
 487                 uint32_t packedDate[2];
 488
 489                 parse_commit(*list);
 490                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 491
 492                 parent = (*list)->parents;
 493
 494                 if (!parent)
 495                         edge_value = GRAPH_PARENT_NONE;
 496                 else {
 497                         edge_value = sha1_pos(parent->item->object.oid.hash,
 498                                               commits,
 499                                               nr_commits,
 500                                               commit_to_sha1);
 501
 502                         if (edge_value < 0)
 503                                 BUG("missing parent %s for commit %s",
 504                                     oid_to_hex(&parent->item->object.oid),
 505                                     oid_to_hex(&(*list)->object.oid));
 506                 }
 507
 508                 hashwrite_be32(f, edge_value);
 509
 510                 if (parent)
 511                         parent = parent->next;
 512
 513                 if (!parent)
 514                         edge_value = GRAPH_PARENT_NONE;
 515                 else if (parent->next)
 516                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 517                 else {
 518                         edge_value = sha1_pos(parent->item->object.oid.hash,
 519                                               commits,
 520                                               nr_commits,
 521                                               commit_to_sha1);
 522                         if (edge_value < 0)
 523                                 BUG("missing parent %s for commit %s",
 524                                     oid_to_hex(&parent->item->object.oid),
 525                                     oid_to_hex(&(*list)->object.oid));
 526                 }
 527
 528                 hashwrite_be32(f, edge_value);
 529
 530                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 531                         do {
 532                                 num_extra_edges++;
 533                                 parent = parent->next;
 534                         } while (parent);
 535                 }
 536
 537                 if (sizeof((*list)->date) > 4)
 538                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 539                 else
 540                         packedDate[0] = 0;
 541
 542                 packedDate[0] |= htonl((*list)->generation << 2);
 543
 544                 packedDate[1] = htonl((*list)->date);
 545                 hashwrite(f, packedDate, 8);
 546
 547                 list++;
 548         }
 549 }
 550
 551 static void write_graph_chunk_large_edges(struct hashfile *f,
 552                                           struct commit **commits,
 553                                           int nr_commits)
 554 {
 555         struct commit **list = commits;
 556         struct commit **last = commits + nr_commits;
 557         struct commit_list *parent;
 558
 559         while (list < last) {
 560                 int num_parents = 0;
 561                 for (parent = (*list)->parents; num_parents < 3 && parent;
 562                      parent = parent->next)
 563                         num_parents++;
 564
 565                 if (num_parents <= 2) {
 566                         list++;
 567                         continue;
 568                 }
 569
 570                 /* Since num_parents > 2, this initializer is safe. */
 571                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 572                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 573                                                   commits,
 574                                                   nr_commits,
 575                                                   commit_to_sha1);
 576
 577                         if (edge_value < 0)
 578                                 BUG("missing parent %s for commit %s",
 579                                     oid_to_hex(&parent->item->object.oid),
 580                                     oid_to_hex(&(*list)->object.oid));
 581                         else if (!parent->next)
 582                                 edge_value |= GRAPH_LAST_EDGE;
 583
 584                         hashwrite_be32(f, edge_value);
 585                 }
 586
 587                 list++;
 588         }
 589 }
 590
 591 static int commit_compare(const void *_a, const void *_b)
 592 {
 593         const struct object_id *a = (const struct object_id *)_a;
 594         const struct object_id *b = (const struct object_id *)_b;
 595         return oidcmp(a, b);
 596 }
 597
 598 struct packed_commit_list {
 599         struct commit **list;
 600         int nr;
 601         int alloc;
 602 };
 603
 604 struct packed_oid_list {
 605         struct object_id *list;
 606         int nr;
 607         int alloc;
 608         struct progress *progress;
 609         int progress_done;
 610 };
 611
 612 static int add_packed_commits(const struct object_id *oid,
 613                               struct packed_git *pack,
 614                               uint32_t pos,
 615                               void *data)
 616 {
 617         struct packed_oid_list *list = (struct packed_oid_list*)data;
 618         enum object_type type;
 619         off_t offset = nth_packed_object_offset(pack, pos);
 620         struct object_info oi = OBJECT_INFO_INIT;
 621
 622         if (list->progress)
 623                 display_progress(list->progress, ++list->progress_done);
 624
 625         oi.typep = &type;
 626         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 627                 die(_("unable to get type of object %s"), oid_to_hex(oid));
 628
 629         if (type != OBJ_COMMIT)
 630                 return 0;
 631
 632         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 633         oidcpy(&(list->list[list->nr]), oid);
 634         list->nr++;
 635
 636         return 0;
 637 }
 638
 639 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 640 {
 641         struct commit_list *parent;
 642         for (parent = commit->parents; parent; parent = parent->next) {
 643                 if (!(parent->item->object.flags & UNINTERESTING)) {
 644                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 645                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 646                         oids->nr++;
 647                         parent->item->object.flags |= UNINTERESTING;
 648                 }
 649         }
 650 }
 651
 652 static void close_reachable(struct packed_oid_list *oids, int report_progress)
 653 {
 654         int i, j;
 655         struct commit *commit;
 656         struct progress *progress = NULL;
 657
 658         if (report_progress)
 659                 progress = start_delayed_progress(
 660                         _("Loading known commits in commit graph"), j = 0);
 661         for (i = 0; i < oids->nr; i++) {
 662                 display_progress(progress, ++j);
 663                 commit = lookup_commit(the_repository, &oids->list[i]);
 664                 if (commit)
 665                         commit->object.flags |= UNINTERESTING;
 666         }
 667         stop_progress(&progress);
 668
 669         /*
 670          * As this loop runs, oids->nr may grow, but not more
 671          * than the number of missing commits in the reachable
 672          * closure.
 673          */
 674         if (report_progress)
 675                 progress = start_delayed_progress(
 676                         _("Expanding reachable commits in commit graph"), j = 0);
 677         for (i = 0; i < oids->nr; i++) {
 678                 display_progress(progress, ++j);
 679                 commit = lookup_commit(the_repository, &oids->list[i]);
 680
 681                 if (commit && !parse_commit(commit))
 682                         add_missing_parents(oids, commit);
 683         }
 684         stop_progress(&progress);
 685
 686         if (report_progress)
 687                 progress = start_delayed_progress(
 688                         _("Clearing commit marks in commit graph"), j = 0);
 689         for (i = 0; i < oids->nr; i++) {
 690                 display_progress(progress, ++j);
 691                 commit = lookup_commit(the_repository, &oids->list[i]);
 692
 693                 if (commit)
 694                         commit->object.flags &= ~UNINTERESTING;
 695         }
 696         stop_progress(&progress);
 697 }
 698
 699 static void compute_generation_numbers(struct packed_commit_list* commits,
 700                                        int report_progress)
 701 {
 702         int i;
 703         struct commit_list *list = NULL;
 704         struct progress *progress = NULL;
 705
 706         if (report_progress)
 707                 progress = start_progress(
 708                         _("Computing commit graph generation numbers"),
 709                         commits->nr);
 710         for (i = 0; i < commits->nr; i++) {
 711                 display_progress(progress, i + 1);
 712                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 713                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 714                         continue;
 715
 716                 commit_list_insert(commits->list[i], &list);
 717                 while (list) {
 718                         struct commit *current = list->item;
 719                         struct commit_list *parent;
 720                         int all_parents_computed = 1;
 721                         uint32_t max_generation = 0;
 722
 723                         for (parent = current->parents; parent; parent = parent->next) {
 724                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 725                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 726                                         all_parents_computed = 0;
 727                                         commit_list_insert(parent->item, &list);
 728                                         break;
 729                                 } else if (parent->item->generation > max_generation) {
 730                                         max_generation = parent->item->generation;
 731                                 }
 732                         }
 733
 734                         if (all_parents_computed) {
 735                                 current->generation = max_generation + 1;
 736                                 pop_commit(&list);
 737
 738                                 if (current->generation > GENERATION_NUMBER_MAX)
 739                                         current->generation = GENERATION_NUMBER_MAX;
 740                         }
 741                 }
 742         }
 743         stop_progress(&progress);
 744 }
 745
 746 static int add_ref_to_list(const char *refname,
 747                            const struct object_id *oid,
 748                            int flags, void *cb_data)
 749 {
 750         struct string_list *list = (struct string_list *)cb_data;
 751
 752         string_list_append(list, oid_to_hex(oid));
 753         return 0;
 754 }
 755
 756 void write_commit_graph_reachable(const char *obj_dir, int append,
 757                                   int report_progress)
 758 {
 759         struct string_list list = STRING_LIST_INIT_DUP;
 760
 761         for_each_ref(add_ref_to_list, &list);
 762         write_commit_graph(obj_dir, NULL, &list, append, report_progress);
 763
 764         string_list_clear(&list, 0);
 765 }
 766
 767 void write_commit_graph(const char *obj_dir,
 768                         struct string_list *pack_indexes,
 769                         struct string_list *commit_hex,
 770                         int append, int report_progress)
 771 {
 772         struct packed_oid_list oids;
 773         struct packed_commit_list commits;
 774         struct hashfile *f;
 775         uint32_t i, count_distinct = 0;
 776         char *graph_name;
 777         struct lock_file lk = LOCK_INIT;
 778         uint32_t chunk_ids[5];
 779         uint64_t chunk_offsets[5];
 780         int num_chunks;
 781         int num_extra_edges;
 782         struct commit_list *parent;
 783         struct progress *progress = NULL;
 784         const unsigned hashsz = the_hash_algo->rawsz;
 785
 786         if (!commit_graph_compatible(the_repository))
 787                 return;
 788
 789         oids.nr = 0;
 790         oids.alloc = approximate_object_count() / 32;
 791         oids.progress = NULL;
 792         oids.progress_done = 0;
 793
 794         if (append) {
 795                 prepare_commit_graph_one(the_repository, obj_dir);
 796                 if (the_repository->objects->commit_graph)
 797                         oids.alloc += the_repository->objects->commit_graph->num_commits;
 798         }
 799
 800         if (oids.alloc < 1024)
 801                 oids.alloc = 1024;
 802         ALLOC_ARRAY(oids.list, oids.alloc);
 803
 804         if (append && the_repository->objects->commit_graph) {
 805                 struct commit_graph *commit_graph =
 806                         the_repository->objects->commit_graph;
 807                 for (i = 0; i < commit_graph->num_commits; i++) {
 808                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 809                                 commit_graph->hash_len * i;
 810                         hashcpy(oids.list[oids.nr++].hash, hash);
 811                 }
 812         }
 813
 814         if (pack_indexes) {
 815                 struct strbuf packname = STRBUF_INIT;
 816                 int dirlen;
 817                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 818                 dirlen = packname.len;
 819                 if (report_progress) {
 820                         oids.progress = start_delayed_progress(
 821                                 _("Finding commits for commit graph"), 0);
 822                         oids.progress_done = 0;
 823                 }
 824                 for (i = 0; i < pack_indexes->nr; i++) {
 825                         struct packed_git *p;
 826                         strbuf_setlen(&packname, dirlen);
 827                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 828                         p = add_packed_git(packname.buf, packname.len, 1);
 829                         if (!p)
 830                                 die(_("error adding pack %s"), packname.buf);
 831                         if (open_pack_index(p))
 832                                 die(_("error opening index for %s"), packname.buf);
 833                         for_each_object_in_pack(p, add_packed_commits, &oids, 0);
 834                         close_pack(p);
 835                         free(p);
 836                 }
 837                 stop_progress(&oids.progress);
 838                 strbuf_release(&packname);
 839         }
 840
 841         if (commit_hex) {
 842                 if (report_progress)
 843                         progress = start_delayed_progress(
 844                                 _("Finding commits for commit graph"),
 845                                 commit_hex->nr);
 846                 for (i = 0; i < commit_hex->nr; i++) {
 847                         const char *end;
 848                         struct object_id oid;
 849                         struct commit *result;
 850
 851                         display_progress(progress, i + 1);
 852                         if (commit_hex->items[i].string &&
 853                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 854                                 continue;
 855
 856                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
 857
 858                         if (result) {
 859                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 860                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 861                                 oids.nr++;
 862                         }
 863                 }
 864                 stop_progress(&progress);
 865         }
 866
 867         if (!pack_indexes && !commit_hex) {
 868                 if (report_progress)
 869                         oids.progress = start_delayed_progress(
 870                                 _("Finding commits for commit graph"), 0);
 871                 for_each_packed_object(add_packed_commits, &oids, 0);
 872                 stop_progress(&oids.progress);
 873         }
 874
 875         close_reachable(&oids, report_progress);
 876
 877         QSORT(oids.list, oids.nr, commit_compare);
 878
 879         count_distinct = 1;
 880         for (i = 1; i < oids.nr; i++) {
 881                 if (!oideq(&oids.list[i - 1], &oids.list[i]))
 882                         count_distinct++;
 883         }
 884
 885         if (count_distinct >= GRAPH_EDGE_LAST_MASK)
 886                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 887
 888         commits.nr = 0;
 889         commits.alloc = count_distinct;
 890         ALLOC_ARRAY(commits.list, commits.alloc);
 891
 892         num_extra_edges = 0;
 893         for (i = 0; i < oids.nr; i++) {
 894                 int num_parents = 0;
 895                 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
 896                         continue;
 897
 898                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
 899                 parse_commit(commits.list[commits.nr]);
 900
 901                 for (parent = commits.list[commits.nr]->parents;
 902                      parent; parent = parent->next)
 903                         num_parents++;
 904
 905                 if (num_parents > 2)
 906                         num_extra_edges += num_parents - 1;
 907
 908                 commits.nr++;
 909         }
 910         num_chunks = num_extra_edges ? 4 : 3;
 911
 912         if (commits.nr >= GRAPH_EDGE_LAST_MASK)
 913                 die(_("too many commits to write graph"));
 914
 915         compute_generation_numbers(&commits, report_progress);
 916
 917         graph_name = get_commit_graph_filename(obj_dir);
 918         if (safe_create_leading_directories(graph_name)) {
 919                 UNLEAK(graph_name);
 920                 die_errno(_("unable to create leading directories of %s"),
 921                           graph_name);
 922         }
 923
 924         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 925         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 926
 927         hashwrite_be32(f, GRAPH_SIGNATURE);
 928
 929         hashwrite_u8(f, GRAPH_VERSION);
 930         hashwrite_u8(f, oid_version());
 931         hashwrite_u8(f, num_chunks);
 932         hashwrite_u8(f, 0); /* unused padding byte */
 933
 934         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 935         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 936         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 937         if (num_extra_edges)
 938                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 939         else
 940                 chunk_ids[3] = 0;
 941         chunk_ids[4] = 0;
 942
 943         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 944         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 945         chunk_offsets[2] = chunk_offsets[1] + hashsz * commits.nr;
 946         chunk_offsets[3] = chunk_offsets[2] + (hashsz + 16) * commits.nr;
 947         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 948
 949         for (i = 0; i <= num_chunks; i++) {
 950                 uint32_t chunk_write[3];
 951
 952                 chunk_write[0] = htonl(chunk_ids[i]);
 953                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 954                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 955                 hashwrite(f, chunk_write, 12);
 956         }
 957
 958         write_graph_chunk_fanout(f, commits.list, commits.nr);
 959         write_graph_chunk_oids(f, hashsz, commits.list, commits.nr);
 960         write_graph_chunk_data(f, hashsz, commits.list, commits.nr);
 961         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 962
 963         close_commit_graph(the_repository);
 964         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 965         commit_lock_file(&lk);
 966
 967         free(graph_name);
 968         free(commits.list);
 969         free(oids.list);
 970 }
 971
 972 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 973 static int verify_commit_graph_error;
 974
 975 static void graph_report(const char *fmt, ...)
 976 {
 977         va_list ap;
 978
 979         verify_commit_graph_error = 1;
 980         va_start(ap, fmt);
 981         vfprintf(stderr, fmt, ap);
 982         fprintf(stderr, "\n");
 983         va_end(ap);
 984 }
 985
 986 #define GENERATION_ZERO_EXISTS 1
 987 #define GENERATION_NUMBER_EXISTS 2
 988
 989 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 990 {
 991         uint32_t i, cur_fanout_pos = 0;
 992         struct object_id prev_oid, cur_oid, checksum;
 993         int generation_zero = 0;
 994         struct hashfile *f;
 995         int devnull;
 996         struct progress *progress = NULL;
 997
 998         if (!g) {
 999                 graph_report("no commit-graph file loaded");
1000                 return 1;
1001         }
1002
1003         verify_commit_graph_error = 0;
1004
1005         if (!g->chunk_oid_fanout)
1006                 graph_report("commit-graph is missing the OID Fanout chunk");
1007         if (!g->chunk_oid_lookup)
1008                 graph_report("commit-graph is missing the OID Lookup chunk");
1009         if (!g->chunk_commit_data)
1010                 graph_report("commit-graph is missing the Commit Data chunk");
1011
1012         if (verify_commit_graph_error)
1013                 return verify_commit_graph_error;
1014
1015         devnull = open("/dev/null", O_WRONLY);
1016         f = hashfd(devnull, NULL);
1017         hashwrite(f, g->data, g->data_len - g->hash_len);
1018         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1019         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1020                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1021                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1022         }
1023
1024         for (i = 0; i < g->num_commits; i++) {
1025                 struct commit *graph_commit;
1026
1027                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1028
1029                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1030                         graph_report("commit-graph has incorrect OID order: %s then %s",
1031                                      oid_to_hex(&prev_oid),
1032                                      oid_to_hex(&cur_oid));
1033
1034                 oidcpy(&prev_oid, &cur_oid);
1035
1036                 while (cur_oid.hash[0] > cur_fanout_pos) {
1037                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1038
1039                         if (i != fanout_value)
1040                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1041                                              cur_fanout_pos, fanout_value, i);
1042                         cur_fanout_pos++;
1043                 }
1044
1045                 graph_commit = lookup_commit(r, &cur_oid);
1046                 if (!parse_commit_in_graph_one(r, g, graph_commit))
1047                         graph_report("failed to parse %s from commit-graph",
1048                                      oid_to_hex(&cur_oid));
1049         }
1050
1051         while (cur_fanout_pos < 256) {
1052                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1053
1054                 if (g->num_commits != fanout_value)
1055                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1056                                      cur_fanout_pos, fanout_value, i);
1057
1058                 cur_fanout_pos++;
1059         }
1060
1061         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1062                 return verify_commit_graph_error;
1063
1064         progress = start_progress(_("Verifying commits in commit graph"),
1065                                   g->num_commits);
1066         for (i = 0; i < g->num_commits; i++) {
1067                 struct commit *graph_commit, *odb_commit;
1068                 struct commit_list *graph_parents, *odb_parents;
1069                 uint32_t max_generation = 0;
1070
1071                 display_progress(progress, i + 1);
1072                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1073
1074                 graph_commit = lookup_commit(r, &cur_oid);
1075                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1076                 if (parse_commit_internal(odb_commit, 0, 0)) {
1077                         graph_report("failed to parse %s from object database",
1078                                      oid_to_hex(&cur_oid));
1079                         continue;
1080                 }
1081
1082                 if (!oideq(&get_commit_tree_in_graph_one(r, g, graph_commit)->object.oid,
1083                            get_commit_tree_oid(odb_commit)))
1084                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1085                                      oid_to_hex(&cur_oid),
1086                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1087                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1088
1089                 graph_parents = graph_commit->parents;
1090                 odb_parents = odb_commit->parents;
1091
1092                 while (graph_parents) {
1093                         if (odb_parents == NULL) {
1094                                 graph_report("commit-graph parent list for commit %s is too long",
1095                                              oid_to_hex(&cur_oid));
1096                                 break;
1097                         }
1098
1099                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1100                                 graph_report("commit-graph parent for %s is %s != %s",
1101                                              oid_to_hex(&cur_oid),
1102                                              oid_to_hex(&graph_parents->item->object.oid),
1103                                              oid_to_hex(&odb_parents->item->object.oid));
1104
1105                         if (graph_parents->item->generation > max_generation)
1106                                 max_generation = graph_parents->item->generation;
1107
1108                         graph_parents = graph_parents->next;
1109                         odb_parents = odb_parents->next;
1110                 }
1111
1112                 if (odb_parents != NULL)
1113                         graph_report("commit-graph parent list for commit %s terminates early",
1114                                      oid_to_hex(&cur_oid));
1115
1116                 if (!graph_commit->generation) {
1117                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1118                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1119                                              oid_to_hex(&cur_oid));
1120                         generation_zero = GENERATION_ZERO_EXISTS;
1121                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1122                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1123                                      oid_to_hex(&cur_oid));
1124
1125                 if (generation_zero == GENERATION_ZERO_EXISTS)
1126                         continue;
1127
1128                 /*
1129                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1130                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1131                  * extra logic in the following condition.
1132                  */
1133                 if (max_generation == GENERATION_NUMBER_MAX)
1134                         max_generation--;
1135
1136                 if (graph_commit->generation != max_generation + 1)
1137                         graph_report("commit-graph generation for commit %s is %u != %u",
1138                                      oid_to_hex(&cur_oid),
1139                                      graph_commit->generation,
1140                                      max_generation + 1);
1141
1142                 if (graph_commit->date != odb_commit->date)
1143                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1144                                      oid_to_hex(&cur_oid),
1145                                      graph_commit->date,
1146                                      odb_commit->date);
1147         }
1148         stop_progress(&progress);
1149
1150         return verify_commit_graph_error;
1151 }
1152
1153 void free_commit_graph(struct commit_graph *g)
1154 {
1155         if (!g)
1156                 return;
1157         if (g->graph_fd >= 0) {
1158                 munmap((void *)g->data, g->data_len);
1159                 g->data = NULL;
1160                 close(g->graph_fd);
1161         }
1162         free(g);
1163 }