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