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