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

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "dir.h"
   4 #include "git-compat-util.h"
   5 #include "lockfile.h"
   6 #include "pack.h"
   7 #include "packfile.h"
   8 #include "commit.h"
   9 #include "object.h"
  10 #include "refs.h"
  11 #include "revision.h"
  12 #include "sha1-lookup.h"
  13 #include "commit-graph.h"
  14 #include "object-store.h"
  15 #include "alloc.h"
  16 #include "hashmap.h"
  17 #include "replace-object.h"
  18 #include "progress.h"
  19
  20 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
  21 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  22 #define GRAPH_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  23 #define GRAPH_CHUNKID_DATA 0x43444154 /* "CDAT" */
  24 #define GRAPH_CHUNKID_LARGEEDGES 0x45444745 /* "EDGE" */
  25
  26 #define GRAPH_DATA_WIDTH 36
  27
  28 #define GRAPH_VERSION_1 0x1
  29 #define GRAPH_VERSION GRAPH_VERSION_1
  30
  31 #define GRAPH_OID_VERSION_SHA1 1
  32 #define GRAPH_OID_LEN_SHA1 GIT_SHA1_RAWSZ
  33 #define GRAPH_OID_VERSION GRAPH_OID_VERSION_SHA1
  34 #define GRAPH_OID_LEN GRAPH_OID_LEN_SHA1
  35
  36 #define GRAPH_OCTOPUS_EDGES_NEEDED 0x80000000
  37 #define GRAPH_PARENT_MISSING 0x7fffffff
  38 #define GRAPH_EDGE_LAST_MASK 0x7fffffff
  39 #define GRAPH_PARENT_NONE 0x70000000
  40
  41 #define GRAPH_LAST_EDGE 0x80000000
  42
  43 #define GRAPH_HEADER_SIZE 8
  44 #define GRAPH_FANOUT_SIZE (4 * 256)
  45 #define GRAPH_CHUNKLOOKUP_WIDTH 12
  46 #define GRAPH_MIN_SIZE (GRAPH_HEADER_SIZE + 4 * GRAPH_CHUNKLOOKUP_WIDTH \
  47                         + GRAPH_FANOUT_SIZE + GRAPH_OID_LEN)
  48
  49 char *get_commit_graph_filename(const char *obj_dir)
  50 {
  51         return xstrfmt("%s/info/commit-graph", obj_dir);
  52 }
  53
  54 static struct commit_graph *alloc_commit_graph(void)
  55 {
  56         struct commit_graph *g = xcalloc(1, sizeof(*g));
  57         g->graph_fd = -1;
  58
  59         return g;
  60 }
  61
  62 extern int read_replace_refs;
  63
  64 static int commit_graph_compatible(struct repository *r)
  65 {
  66         if (!r->gitdir)
  67                 return 0;
  68
  69         if (read_replace_refs) {
  70                 prepare_replace_object(r);
  71                 if (hashmap_get_size(&r->objects->replace_map->map))
  72                         return 0;
  73         }
  74
  75         prepare_commit_graft(r);
  76         if (r->parsed_objects && r->parsed_objects->grafts_nr)
  77                 return 0;
  78         if (is_repository_shallow(r))
  79                 return 0;
  80
  81         return 1;
  82 }
  83
  84 struct commit_graph *load_commit_graph_one(const char *graph_file)
  85 {
  86         void *graph_map;
  87         size_t graph_size;
  88         struct stat st;
  89         struct commit_graph *ret;
  90         int fd = git_open(graph_file);
  91
  92         if (fd < 0)
  93                 return NULL;
  94         if (fstat(fd, &st)) {
  95                 close(fd);
  96                 return NULL;
  97         }
  98         graph_size = xsize_t(st.st_size);
  99
 100         if (graph_size < GRAPH_MIN_SIZE) {
 101                 close(fd);
 102                 die(_("graph file %s is too small"), graph_file);
 103         }
 104         graph_map = xmmap(NULL, graph_size, PROT_READ, MAP_PRIVATE, fd, 0);
 105         ret = parse_commit_graph(graph_map, fd, graph_size);
 106
 107         if (!ret) {
 108                 munmap(graph_map, graph_size);
 109                 close(fd);
 110                 exit(1);
 111         }
 112
 113         return ret;
 114 }
 115
 116 struct commit_graph *parse_commit_graph(void *graph_map, int fd,
 117                                         size_t graph_size)
 118 {
 119         const unsigned char *data, *chunk_lookup;
 120         uint32_t i;
 121         struct commit_graph *graph;
 122         uint64_t last_chunk_offset;
 123         uint32_t last_chunk_id;
 124         uint32_t graph_signature;
 125         unsigned char graph_version, hash_version;
 126
 127         if (!graph_map)
 128                 return NULL;
 129
 130         if (graph_size < GRAPH_MIN_SIZE)
 131                 return NULL;
 132
 133         data = (const unsigned char *)graph_map;
 134
 135         graph_signature = get_be32(data);
 136         if (graph_signature != GRAPH_SIGNATURE) {
 137                 error(_("graph signature %X does not match signature %X"),
 138                       graph_signature, GRAPH_SIGNATURE);
 139                 return NULL;
 140         }
 141
 142         graph_version = *(unsigned char*)(data + 4);
 143         if (graph_version != GRAPH_VERSION) {
 144                 error(_("graph version %X does not match version %X"),
 145                       graph_version, GRAPH_VERSION);
 146                 return NULL;
 147         }
 148
 149         hash_version = *(unsigned char*)(data + 5);
 150         if (hash_version != GRAPH_OID_VERSION) {
 151                 error(_("hash version %X does not match version %X"),
 152                       hash_version, GRAPH_OID_VERSION);
 153                 return NULL;
 154         }
 155
 156         graph = alloc_commit_graph();
 157
 158         graph->hash_len = GRAPH_OID_LEN;
 159         graph->num_chunks = *(unsigned char*)(data + 6);
 160         graph->graph_fd = fd;
 161         graph->data = graph_map;
 162         graph->data_len = graph_size;
 163
 164         last_chunk_id = 0;
 165         last_chunk_offset = 8;
 166         chunk_lookup = data + 8;
 167         for (i = 0; i < graph->num_chunks; i++) {
 168                 uint32_t chunk_id = get_be32(chunk_lookup + 0);
 169                 uint64_t chunk_offset = get_be64(chunk_lookup + 4);
 170                 int chunk_repeated = 0;
 171
 172                 chunk_lookup += GRAPH_CHUNKLOOKUP_WIDTH;
 173
 174                 if (chunk_offset > graph_size - GIT_MAX_RAWSZ) {
 175                         error(_("improper chunk offset %08x%08x"), (uint32_t)(chunk_offset >> 32),
 176                               (uint32_t)chunk_offset);
 177                         free(graph);
 178                         return NULL;
 179                 }
 180
 181                 switch (chunk_id) {
 182                 case GRAPH_CHUNKID_OIDFANOUT:
 183                         if (graph->chunk_oid_fanout)
 184                                 chunk_repeated = 1;
 185                         else
 186                                 graph->chunk_oid_fanout = (uint32_t*)(data + chunk_offset);
 187                         break;
 188
 189                 case GRAPH_CHUNKID_OIDLOOKUP:
 190                         if (graph->chunk_oid_lookup)
 191                                 chunk_repeated = 1;
 192                         else
 193                                 graph->chunk_oid_lookup = data + chunk_offset;
 194                         break;
 195
 196                 case GRAPH_CHUNKID_DATA:
 197                         if (graph->chunk_commit_data)
 198                                 chunk_repeated = 1;
 199                         else
 200                                 graph->chunk_commit_data = data + chunk_offset;
 201                         break;
 202
 203                 case GRAPH_CHUNKID_LARGEEDGES:
 204                         if (graph->chunk_large_edges)
 205                                 chunk_repeated = 1;
 206                         else
 207                                 graph->chunk_large_edges = data + chunk_offset;
 208                         break;
 209                 }
 210
 211                 if (chunk_repeated) {
 212                         error(_("chunk id %08x appears multiple times"), chunk_id);
 213                         free(graph);
 214                         return NULL;
 215                 }
 216
 217                 if (last_chunk_id == GRAPH_CHUNKID_OIDLOOKUP)
 218                 {
 219                         graph->num_commits = (chunk_offset - last_chunk_offset)
 220                                              / graph->hash_len;
 221                 }
 222
 223                 last_chunk_id = chunk_id;
 224                 last_chunk_offset = chunk_offset;
 225         }
 226
 227         return graph;
 228 }
 229
 230 static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 231 {
 232         char *graph_name;
 233
 234         if (r->objects->commit_graph)
 235                 return;
 236
 237         graph_name = get_commit_graph_filename(obj_dir);
 238         r->objects->commit_graph =
 239                 load_commit_graph_one(graph_name);
 240
 241         FREE_AND_NULL(graph_name);
 242 }
 243
 244 /*
 245  * Return 1 if commit_graph is non-NULL, and 0 otherwise.
 246  *
 247  * On the first invocation, this function attemps to load the commit
 248  * graph if the_repository is configured to have one.
 249  */
 250 static int prepare_commit_graph(struct repository *r)
 251 {
 252         struct object_directory *odb;
 253         int config_value;
 254
 255         if (r->objects->commit_graph_attempted)
 256                 return !!r->objects->commit_graph;
 257         r->objects->commit_graph_attempted = 1;
 258
 259         if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
 260             (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 261             !config_value))
 262                 /*
 263                  * This repository is not configured to use commit graphs, so
 264                  * do not load one. (But report commit_graph_attempted anyway
 265                  * so that commit graph loading is not attempted again for this
 266                  * repository.)
 267                  */
 268                 return 0;
 269
 270         if (!commit_graph_compatible(r))
 271                 return 0;
 272
 273         prepare_alt_odb(r);
 274         for (odb = r->objects->odb;
 275              !r->objects->commit_graph && odb;
 276              odb = odb->next)
 277                 prepare_commit_graph_one(r, odb->path);
 278         return !!r->objects->commit_graph;
 279 }
 280
 281 int generation_numbers_enabled(struct repository *r)
 282 {
 283         uint32_t first_generation;
 284         struct commit_graph *g;
 285         if (!prepare_commit_graph(r))
 286                return 0;
 287
 288         g = r->objects->commit_graph;
 289
 290         if (!g->num_commits)
 291                 return 0;
 292
 293         first_generation = get_be32(g->chunk_commit_data +
 294                                     g->hash_len + 8) >> 2;
 295
 296         return !!first_generation;
 297 }
 298
 299 void close_commit_graph(struct repository *r)
 300 {
 301         free_commit_graph(r->objects->commit_graph);
 302         r->objects->commit_graph = NULL;
 303 }
 304
 305 static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 306 {
 307         return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 308                             g->chunk_oid_lookup, g->hash_len, pos);
 309 }
 310
 311 static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 312                                                  uint64_t pos,
 313                                                  struct commit_list **pptr)
 314 {
 315         struct commit *c;
 316         struct object_id oid;
 317
 318         if (pos >= g->num_commits)
 319                 die("invalid parent position %"PRIu64, pos);
 320
 321         hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 322         c = lookup_commit(the_repository, &oid);
 323         if (!c)
 324                 die(_("could not find commit %s"), oid_to_hex(&oid));
 325         c->graph_pos = pos;
 326         return &commit_list_insert(c, pptr)->next;
 327 }
 328
 329 static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 330 {
 331         const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 332         item->graph_pos = pos;
 333         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 334 }
 335
 336 static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 337 {
 338         uint32_t edge_value;
 339         uint32_t *parent_data_ptr;
 340         uint64_t date_low, date_high;
 341         struct commit_list **pptr;
 342         const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 343
 344         item->object.parsed = 1;
 345         item->graph_pos = pos;
 346
 347         item->maybe_tree = NULL;
 348
 349         date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 350         date_low = get_be32(commit_data + g->hash_len + 12);
 351         item->date = (timestamp_t)((date_high << 32) | date_low);
 352
 353         item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 354
 355         pptr = &item->parents;
 356
 357         edge_value = get_be32(commit_data + g->hash_len);
 358         if (edge_value == GRAPH_PARENT_NONE)
 359                 return 1;
 360         pptr = insert_parent_or_die(g, edge_value, pptr);
 361
 362         edge_value = get_be32(commit_data + g->hash_len + 4);
 363         if (edge_value == GRAPH_PARENT_NONE)
 364                 return 1;
 365         if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 366                 pptr = insert_parent_or_die(g, edge_value, pptr);
 367                 return 1;
 368         }
 369
 370         parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 371                           4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 372         do {
 373                 edge_value = get_be32(parent_data_ptr);
 374                 pptr = insert_parent_or_die(g,
 375                                             edge_value & GRAPH_EDGE_LAST_MASK,
 376                                             pptr);
 377                 parent_data_ptr++;
 378         } while (!(edge_value & GRAPH_LAST_EDGE));
 379
 380         return 1;
 381 }
 382
 383 static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 384 {
 385         if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 386                 *pos = item->graph_pos;
 387                 return 1;
 388         } else {
 389                 return bsearch_graph(g, &(item->object.oid), pos);
 390         }
 391 }
 392
 393 static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 394 {
 395         uint32_t pos;
 396
 397         if (item->object.parsed)
 398                 return 1;
 399
 400         if (find_commit_in_graph(item, g, &pos))
 401                 return fill_commit_in_graph(item, g, pos);
 402
 403         return 0;
 404 }
 405
 406 int parse_commit_in_graph(struct repository *r, struct commit *item)
 407 {
 408         if (!prepare_commit_graph(r))
 409                 return 0;
 410         return parse_commit_in_graph_one(r->objects->commit_graph, item);
 411 }
 412
 413 void load_commit_graph_info(struct repository *r, struct commit *item)
 414 {
 415         uint32_t pos;
 416         if (!prepare_commit_graph(r))
 417                 return;
 418         if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 419                 fill_commit_graph_info(item, r->objects->commit_graph, pos);
 420 }
 421
 422 static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 423 {
 424         struct object_id oid;
 425         const unsigned char *commit_data = g->chunk_commit_data +
 426                                            GRAPH_DATA_WIDTH * (c->graph_pos);
 427
 428         hashcpy(oid.hash, commit_data);
 429         c->maybe_tree = lookup_tree(the_repository, &oid);
 430
 431         return c->maybe_tree;
 432 }
 433
 434 static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 435                                                  const struct commit *c)
 436 {
 437         if (c->maybe_tree)
 438                 return c->maybe_tree;
 439         if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 440                 BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 441
 442         return load_tree_for_commit(g, (struct commit *)c);
 443 }
 444
 445 struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 446 {
 447         return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
 448 }
 449
 450 static void write_graph_chunk_fanout(struct hashfile *f,
 451                                      struct commit **commits,
 452                                      int nr_commits)
 453 {
 454         int i, count = 0;
 455         struct commit **list = commits;
 456
 457         /*
 458          * Write the first-level table (the list is sorted,
 459          * but we use a 256-entry lookup to be able to avoid
 460          * having to do eight extra binary search iterations).
 461          */
 462         for (i = 0; i < 256; i++) {
 463                 while (count < nr_commits) {
 464                         if ((*list)->object.oid.hash[0] != i)
 465                                 break;
 466                         count++;
 467                         list++;
 468                 }
 469
 470                 hashwrite_be32(f, count);
 471         }
 472 }
 473
 474 static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 475                                    struct commit **commits, int nr_commits)
 476 {
 477         struct commit **list = commits;
 478         int count;
 479         for (count = 0; count < nr_commits; count++, list++)
 480                 hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 481 }
 482
 483 static const unsigned char *commit_to_sha1(size_t index, void *table)
 484 {
 485         struct commit **commits = table;
 486         return commits[index]->object.oid.hash;
 487 }
 488
 489 static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 490                                    struct commit **commits, int nr_commits)
 491 {
 492         struct commit **list = commits;
 493         struct commit **last = commits + nr_commits;
 494         uint32_t num_extra_edges = 0;
 495
 496         while (list < last) {
 497                 struct commit_list *parent;
 498                 int edge_value;
 499                 uint32_t packedDate[2];
 500
 501                 parse_commit(*list);
 502                 hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 503
 504                 parent = (*list)->parents;
 505
 506                 if (!parent)
 507                         edge_value = GRAPH_PARENT_NONE;
 508                 else {
 509                         edge_value = sha1_pos(parent->item->object.oid.hash,
 510                                               commits,
 511                                               nr_commits,
 512                                               commit_to_sha1);
 513
 514                         if (edge_value < 0)
 515                                 edge_value = GRAPH_PARENT_MISSING;
 516                 }
 517
 518                 hashwrite_be32(f, edge_value);
 519
 520                 if (parent)
 521                         parent = parent->next;
 522
 523                 if (!parent)
 524                         edge_value = GRAPH_PARENT_NONE;
 525                 else if (parent->next)
 526                         edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 527                 else {
 528                         edge_value = sha1_pos(parent->item->object.oid.hash,
 529                                               commits,
 530                                               nr_commits,
 531                                               commit_to_sha1);
 532                         if (edge_value < 0)
 533                                 edge_value = GRAPH_PARENT_MISSING;
 534                 }
 535
 536                 hashwrite_be32(f, edge_value);
 537
 538                 if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 539                         do {
 540                                 num_extra_edges++;
 541                                 parent = parent->next;
 542                         } while (parent);
 543                 }
 544
 545                 if (sizeof((*list)->date) > 4)
 546                         packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 547                 else
 548                         packedDate[0] = 0;
 549
 550                 packedDate[0] |= htonl((*list)->generation << 2);
 551
 552                 packedDate[1] = htonl((*list)->date);
 553                 hashwrite(f, packedDate, 8);
 554
 555                 list++;
 556         }
 557 }
 558
 559 static void write_graph_chunk_large_edges(struct hashfile *f,
 560                                           struct commit **commits,
 561                                           int nr_commits)
 562 {
 563         struct commit **list = commits;
 564         struct commit **last = commits + nr_commits;
 565         struct commit_list *parent;
 566
 567         while (list < last) {
 568                 int num_parents = 0;
 569                 for (parent = (*list)->parents; num_parents < 3 && parent;
 570                      parent = parent->next)
 571                         num_parents++;
 572
 573                 if (num_parents <= 2) {
 574                         list++;
 575                         continue;
 576                 }
 577
 578                 /* Since num_parents > 2, this initializer is safe. */
 579                 for (parent = (*list)->parents->next; parent; parent = parent->next) {
 580                         int edge_value = sha1_pos(parent->item->object.oid.hash,
 581                                                   commits,
 582                                                   nr_commits,
 583                                                   commit_to_sha1);
 584
 585                         if (edge_value < 0)
 586                                 edge_value = GRAPH_PARENT_MISSING;
 587                         else if (!parent->next)
 588                                 edge_value |= GRAPH_LAST_EDGE;
 589
 590                         hashwrite_be32(f, edge_value);
 591                 }
 592
 593                 list++;
 594         }
 595 }
 596
 597 static int commit_compare(const void *_a, const void *_b)
 598 {
 599         const struct object_id *a = (const struct object_id *)_a;
 600         const struct object_id *b = (const struct object_id *)_b;
 601         return oidcmp(a, b);
 602 }
 603
 604 struct packed_commit_list {
 605         struct commit **list;
 606         int nr;
 607         int alloc;
 608 };
 609
 610 struct packed_oid_list {
 611         struct object_id *list;
 612         int nr;
 613         int alloc;
 614         struct progress *progress;
 615         int progress_done;
 616 };
 617
 618 static int add_packed_commits(const struct object_id *oid,
 619                               struct packed_git *pack,
 620                               uint32_t pos,
 621                               void *data)
 622 {
 623         struct packed_oid_list *list = (struct packed_oid_list*)data;
 624         enum object_type type;
 625         off_t offset = nth_packed_object_offset(pack, pos);
 626         struct object_info oi = OBJECT_INFO_INIT;
 627
 628         if (list->progress)
 629                 display_progress(list->progress, ++list->progress_done);
 630
 631         oi.typep = &type;
 632         if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 633                 die(_("unable to get type of object %s"), oid_to_hex(oid));
 634
 635         if (type != OBJ_COMMIT)
 636                 return 0;
 637
 638         ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 639         oidcpy(&(list->list[list->nr]), oid);
 640         list->nr++;
 641
 642         return 0;
 643 }
 644
 645 static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 646 {
 647         struct commit_list *parent;
 648         for (parent = commit->parents; parent; parent = parent->next) {
 649                 if (!(parent->item->object.flags & UNINTERESTING)) {
 650                         ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 651                         oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 652                         oids->nr++;
 653                         parent->item->object.flags |= UNINTERESTING;
 654                 }
 655         }
 656 }
 657
 658 static void close_reachable(struct packed_oid_list *oids, int report_progress)
 659 {
 660         int i, j;
 661         struct commit *commit;
 662         struct progress *progress = NULL;
 663
 664         if (report_progress)
 665                 progress = start_delayed_progress(
 666                         _("Loading known commits in commit graph"), j = 0);
 667         for (i = 0; i < oids->nr; i++) {
 668                 display_progress(progress, ++j);
 669                 commit = lookup_commit(the_repository, &oids->list[i]);
 670                 if (commit)
 671                         commit->object.flags |= UNINTERESTING;
 672         }
 673         stop_progress(&progress);
 674
 675         /*
 676          * As this loop runs, oids->nr may grow, but not more
 677          * than the number of missing commits in the reachable
 678          * closure.
 679          */
 680         if (report_progress)
 681                 progress = start_delayed_progress(
 682                         _("Expanding reachable commits in commit graph"), j = 0);
 683         for (i = 0; i < oids->nr; i++) {
 684                 display_progress(progress, ++j);
 685                 commit = lookup_commit(the_repository, &oids->list[i]);
 686
 687                 if (commit && !parse_commit(commit))
 688                         add_missing_parents(oids, commit);
 689         }
 690         stop_progress(&progress);
 691
 692         if (report_progress)
 693                 progress = start_delayed_progress(
 694                         _("Clearing commit marks in commit graph"), j = 0);
 695         for (i = 0; i < oids->nr; i++) {
 696                 display_progress(progress, ++j);
 697                 commit = lookup_commit(the_repository, &oids->list[i]);
 698
 699                 if (commit)
 700                         commit->object.flags &= ~UNINTERESTING;
 701         }
 702         stop_progress(&progress);
 703 }
 704
 705 static void compute_generation_numbers(struct packed_commit_list* commits,
 706                                        int report_progress)
 707 {
 708         int i;
 709         struct commit_list *list = NULL;
 710         struct progress *progress = NULL;
 711
 712         if (report_progress)
 713                 progress = start_progress(
 714                         _("Computing commit graph generation numbers"),
 715                         commits->nr);
 716         for (i = 0; i < commits->nr; i++) {
 717                 display_progress(progress, i + 1);
 718                 if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 719                     commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 720                         continue;
 721
 722                 commit_list_insert(commits->list[i], &list);
 723                 while (list) {
 724                         struct commit *current = list->item;
 725                         struct commit_list *parent;
 726                         int all_parents_computed = 1;
 727                         uint32_t max_generation = 0;
 728
 729                         for (parent = current->parents; parent; parent = parent->next) {
 730                                 if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 731                                     parent->item->generation == GENERATION_NUMBER_ZERO) {
 732                                         all_parents_computed = 0;
 733                                         commit_list_insert(parent->item, &list);
 734                                         break;
 735                                 } else if (parent->item->generation > max_generation) {
 736                                         max_generation = parent->item->generation;
 737                                 }
 738                         }
 739
 740                         if (all_parents_computed) {
 741                                 current->generation = max_generation + 1;
 742                                 pop_commit(&list);
 743
 744                                 if (current->generation > GENERATION_NUMBER_MAX)
 745                                         current->generation = GENERATION_NUMBER_MAX;
 746                         }
 747                 }
 748         }
 749         stop_progress(&progress);
 750 }
 751
 752 static int add_ref_to_list(const char *refname,
 753                            const struct object_id *oid,
 754                            int flags, void *cb_data)
 755 {
 756         struct string_list *list = (struct string_list *)cb_data;
 757
 758         string_list_append(list, oid_to_hex(oid));
 759         return 0;
 760 }
 761
 762 void write_commit_graph_reachable(const char *obj_dir, int append,
 763                                   int report_progress)
 764 {
 765         struct string_list list = STRING_LIST_INIT_DUP;
 766
 767         for_each_ref(add_ref_to_list, &list);
 768         write_commit_graph(obj_dir, NULL, &list, append, report_progress);
 769
 770         string_list_clear(&list, 0);
 771 }
 772
 773 void write_commit_graph(const char *obj_dir,
 774                         struct string_list *pack_indexes,
 775                         struct string_list *commit_hex,
 776                         int append, int report_progress)
 777 {
 778         struct packed_oid_list oids;
 779         struct packed_commit_list commits;
 780         struct hashfile *f;
 781         uint32_t i, count_distinct = 0;
 782         char *graph_name;
 783         struct lock_file lk = LOCK_INIT;
 784         uint32_t chunk_ids[5];
 785         uint64_t chunk_offsets[5];
 786         int num_chunks;
 787         int num_extra_edges;
 788         struct commit_list *parent;
 789         struct progress *progress = NULL;
 790
 791         if (!commit_graph_compatible(the_repository))
 792                 return;
 793
 794         oids.nr = 0;
 795         oids.alloc = approximate_object_count() / 32;
 796         oids.progress = NULL;
 797         oids.progress_done = 0;
 798
 799         if (append) {
 800                 prepare_commit_graph_one(the_repository, obj_dir);
 801                 if (the_repository->objects->commit_graph)
 802                         oids.alloc += the_repository->objects->commit_graph->num_commits;
 803         }
 804
 805         if (oids.alloc < 1024)
 806                 oids.alloc = 1024;
 807         ALLOC_ARRAY(oids.list, oids.alloc);
 808
 809         if (append && the_repository->objects->commit_graph) {
 810                 struct commit_graph *commit_graph =
 811                         the_repository->objects->commit_graph;
 812                 for (i = 0; i < commit_graph->num_commits; i++) {
 813                         const unsigned char *hash = commit_graph->chunk_oid_lookup +
 814                                 commit_graph->hash_len * i;
 815                         hashcpy(oids.list[oids.nr++].hash, hash);
 816                 }
 817         }
 818
 819         if (pack_indexes) {
 820                 struct strbuf packname = STRBUF_INIT;
 821                 int dirlen;
 822                 strbuf_addf(&packname, "%s/pack/", obj_dir);
 823                 dirlen = packname.len;
 824                 if (report_progress) {
 825                         oids.progress = start_delayed_progress(
 826                                 _("Finding commits for commit graph"), 0);
 827                         oids.progress_done = 0;
 828                 }
 829                 for (i = 0; i < pack_indexes->nr; i++) {
 830                         struct packed_git *p;
 831                         strbuf_setlen(&packname, dirlen);
 832                         strbuf_addstr(&packname, pack_indexes->items[i].string);
 833                         p = add_packed_git(packname.buf, packname.len, 1);
 834                         if (!p)
 835                                 die(_("error adding pack %s"), packname.buf);
 836                         if (open_pack_index(p))
 837                                 die(_("error opening index for %s"), packname.buf);
 838                         for_each_object_in_pack(p, add_packed_commits, &oids, 0);
 839                         close_pack(p);
 840                         free(p);
 841                 }
 842                 stop_progress(&oids.progress);
 843                 strbuf_release(&packname);
 844         }
 845
 846         if (commit_hex) {
 847                 if (report_progress)
 848                         progress = start_delayed_progress(
 849                                 _("Finding commits for commit graph"),
 850                                 commit_hex->nr);
 851                 for (i = 0; i < commit_hex->nr; i++) {
 852                         const char *end;
 853                         struct object_id oid;
 854                         struct commit *result;
 855
 856                         display_progress(progress, i + 1);
 857                         if (commit_hex->items[i].string &&
 858                             parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 859                                 continue;
 860
 861                         result = lookup_commit_reference_gently(the_repository, &oid, 1);
 862
 863                         if (result) {
 864                                 ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 865                                 oidcpy(&oids.list[oids.nr], &(result->object.oid));
 866                                 oids.nr++;
 867                         }
 868                 }
 869                 stop_progress(&progress);
 870         }
 871
 872         if (!pack_indexes && !commit_hex) {
 873                 if (report_progress)
 874                         oids.progress = start_delayed_progress(
 875                                 _("Finding commits for commit graph"), 0);
 876                 for_each_packed_object(add_packed_commits, &oids, 0);
 877                 stop_progress(&oids.progress);
 878         }
 879
 880         close_reachable(&oids, report_progress);
 881
 882         QSORT(oids.list, oids.nr, commit_compare);
 883
 884         count_distinct = 1;
 885         for (i = 1; i < oids.nr; i++) {
 886                 if (!oideq(&oids.list[i - 1], &oids.list[i]))
 887                         count_distinct++;
 888         }
 889
 890         if (count_distinct >= GRAPH_PARENT_MISSING)
 891                 die(_("the commit graph format cannot write %d commits"), count_distinct);
 892
 893         commits.nr = 0;
 894         commits.alloc = count_distinct;
 895         ALLOC_ARRAY(commits.list, commits.alloc);
 896
 897         num_extra_edges = 0;
 898         for (i = 0; i < oids.nr; i++) {
 899                 int num_parents = 0;
 900                 if (i > 0 && oideq(&oids.list[i - 1], &oids.list[i]))
 901                         continue;
 902
 903                 commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
 904                 parse_commit(commits.list[commits.nr]);
 905
 906                 for (parent = commits.list[commits.nr]->parents;
 907                      parent; parent = parent->next)
 908                         num_parents++;
 909
 910                 if (num_parents > 2)
 911                         num_extra_edges += num_parents - 1;
 912
 913                 commits.nr++;
 914         }
 915         num_chunks = num_extra_edges ? 4 : 3;
 916
 917         if (commits.nr >= GRAPH_PARENT_MISSING)
 918                 die(_("too many commits to write graph"));
 919
 920         compute_generation_numbers(&commits, report_progress);
 921
 922         graph_name = get_commit_graph_filename(obj_dir);
 923         if (safe_create_leading_directories(graph_name)) {
 924                 UNLEAK(graph_name);
 925                 die_errno(_("unable to create leading directories of %s"),
 926                           graph_name);
 927         }
 928
 929         hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 930         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 931
 932         hashwrite_be32(f, GRAPH_SIGNATURE);
 933
 934         hashwrite_u8(f, GRAPH_VERSION);
 935         hashwrite_u8(f, GRAPH_OID_VERSION);
 936         hashwrite_u8(f, num_chunks);
 937         hashwrite_u8(f, 0); /* unused padding byte */
 938
 939         chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 940         chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 941         chunk_ids[2] = GRAPH_CHUNKID_DATA;
 942         if (num_extra_edges)
 943                 chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 944         else
 945                 chunk_ids[3] = 0;
 946         chunk_ids[4] = 0;
 947
 948         chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 949         chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 950         chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 951         chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 952         chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 953
 954         for (i = 0; i <= num_chunks; i++) {
 955                 uint32_t chunk_write[3];
 956
 957                 chunk_write[0] = htonl(chunk_ids[i]);
 958                 chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 959                 chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 960                 hashwrite(f, chunk_write, 12);
 961         }
 962
 963         write_graph_chunk_fanout(f, commits.list, commits.nr);
 964         write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 965         write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 966         write_graph_chunk_large_edges(f, commits.list, commits.nr);
 967
 968         close_commit_graph(the_repository);
 969         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 970         commit_lock_file(&lk);
 971
 972         free(graph_name);
 973         free(commits.list);
 974         free(oids.list);
 975 }
 976
 977 #define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 978 static int verify_commit_graph_error;
 979
 980 static void graph_report(const char *fmt, ...)
 981 {
 982         va_list ap;
 983
 984         verify_commit_graph_error = 1;
 985         va_start(ap, fmt);
 986         vfprintf(stderr, fmt, ap);
 987         fprintf(stderr, "\n");
 988         va_end(ap);
 989 }
 990
 991 #define GENERATION_ZERO_EXISTS 1
 992 #define GENERATION_NUMBER_EXISTS 2
 993
 994 int verify_commit_graph(struct repository *r, struct commit_graph *g)
 995 {
 996         uint32_t i, cur_fanout_pos = 0;
 997         struct object_id prev_oid, cur_oid, checksum;
 998         int generation_zero = 0;
 999         struct hashfile *f;
1000         int devnull;
1001         struct progress *progress = NULL;
1002
1003         if (!g) {
1004                 graph_report("no commit-graph file loaded");
1005                 return 1;
1006         }
1007
1008         verify_commit_graph_error = 0;
1009
1010         if (!g->chunk_oid_fanout)
1011                 graph_report("commit-graph is missing the OID Fanout chunk");
1012         if (!g->chunk_oid_lookup)
1013                 graph_report("commit-graph is missing the OID Lookup chunk");
1014         if (!g->chunk_commit_data)
1015                 graph_report("commit-graph is missing the Commit Data chunk");
1016
1017         if (verify_commit_graph_error)
1018                 return verify_commit_graph_error;
1019
1020         devnull = open("/dev/null", O_WRONLY);
1021         f = hashfd(devnull, NULL);
1022         hashwrite(f, g->data, g->data_len - g->hash_len);
1023         finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
1024         if (!hasheq(checksum.hash, g->data + g->data_len - g->hash_len)) {
1025                 graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
1026                 verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
1027         }
1028
1029         for (i = 0; i < g->num_commits; i++) {
1030                 struct commit *graph_commit;
1031
1032                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1033
1034                 if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
1035                         graph_report("commit-graph has incorrect OID order: %s then %s",
1036                                      oid_to_hex(&prev_oid),
1037                                      oid_to_hex(&cur_oid));
1038
1039                 oidcpy(&prev_oid, &cur_oid);
1040
1041                 while (cur_oid.hash[0] > cur_fanout_pos) {
1042                         uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1043
1044                         if (i != fanout_value)
1045                                 graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1046                                              cur_fanout_pos, fanout_value, i);
1047                         cur_fanout_pos++;
1048                 }
1049
1050                 graph_commit = lookup_commit(r, &cur_oid);
1051                 if (!parse_commit_in_graph_one(g, graph_commit))
1052                         graph_report("failed to parse %s from commit-graph",
1053                                      oid_to_hex(&cur_oid));
1054         }
1055
1056         while (cur_fanout_pos < 256) {
1057                 uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
1058
1059                 if (g->num_commits != fanout_value)
1060                         graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
1061                                      cur_fanout_pos, fanout_value, i);
1062
1063                 cur_fanout_pos++;
1064         }
1065
1066         if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
1067                 return verify_commit_graph_error;
1068
1069         progress = start_progress(_("Verifying commits in commit graph"),
1070                                   g->num_commits);
1071         for (i = 0; i < g->num_commits; i++) {
1072                 struct commit *graph_commit, *odb_commit;
1073                 struct commit_list *graph_parents, *odb_parents;
1074                 uint32_t max_generation = 0;
1075
1076                 display_progress(progress, i + 1);
1077                 hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
1078
1079                 graph_commit = lookup_commit(r, &cur_oid);
1080                 odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
1081                 if (parse_commit_internal(odb_commit, 0, 0)) {
1082                         graph_report("failed to parse %s from object database",
1083                                      oid_to_hex(&cur_oid));
1084                         continue;
1085                 }
1086
1087                 if (!oideq(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
1088                            get_commit_tree_oid(odb_commit)))
1089                         graph_report("root tree OID for commit %s in commit-graph is %s != %s",
1090                                      oid_to_hex(&cur_oid),
1091                                      oid_to_hex(get_commit_tree_oid(graph_commit)),
1092                                      oid_to_hex(get_commit_tree_oid(odb_commit)));
1093
1094                 graph_parents = graph_commit->parents;
1095                 odb_parents = odb_commit->parents;
1096
1097                 while (graph_parents) {
1098                         if (odb_parents == NULL) {
1099                                 graph_report("commit-graph parent list for commit %s is too long",
1100                                              oid_to_hex(&cur_oid));
1101                                 break;
1102                         }
1103
1104                         if (!oideq(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
1105                                 graph_report("commit-graph parent for %s is %s != %s",
1106                                              oid_to_hex(&cur_oid),
1107                                              oid_to_hex(&graph_parents->item->object.oid),
1108                                              oid_to_hex(&odb_parents->item->object.oid));
1109
1110                         if (graph_parents->item->generation > max_generation)
1111                                 max_generation = graph_parents->item->generation;
1112
1113                         graph_parents = graph_parents->next;
1114                         odb_parents = odb_parents->next;
1115                 }
1116
1117                 if (odb_parents != NULL)
1118                         graph_report("commit-graph parent list for commit %s terminates early",
1119                                      oid_to_hex(&cur_oid));
1120
1121                 if (!graph_commit->generation) {
1122                         if (generation_zero == GENERATION_NUMBER_EXISTS)
1123                                 graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1124                                              oid_to_hex(&cur_oid));
1125                         generation_zero = GENERATION_ZERO_EXISTS;
1126                 } else if (generation_zero == GENERATION_ZERO_EXISTS)
1127                         graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1128                                      oid_to_hex(&cur_oid));
1129
1130                 if (generation_zero == GENERATION_ZERO_EXISTS)
1131                         continue;
1132
1133                 /*
1134                  * If one of our parents has generation GENERATION_NUMBER_MAX, then
1135                  * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1136                  * extra logic in the following condition.
1137                  */
1138                 if (max_generation == GENERATION_NUMBER_MAX)
1139                         max_generation--;
1140
1141                 if (graph_commit->generation != max_generation + 1)
1142                         graph_report("commit-graph generation for commit %s is %u != %u",
1143                                      oid_to_hex(&cur_oid),
1144                                      graph_commit->generation,
1145                                      max_generation + 1);
1146
1147                 if (graph_commit->date != odb_commit->date)
1148                         graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1149                                      oid_to_hex(&cur_oid),
1150                                      graph_commit->date,
1151                                      odb_commit->date);
1152         }
1153         stop_progress(&progress);
1154
1155         return verify_commit_graph_error;
1156 }
1157
1158 void free_commit_graph(struct commit_graph *g)
1159 {
1160         if (!g)
1161                 return;
1162         if (g->graph_fd >= 0) {
1163                 munmap((void *)g->data, g->data_len);
1164                 g->data = NULL;
1165                 close(g->graph_fd);
1166         }
1167         free(g);
1168 }