commit-graph.con commit t1406: make hash-size independent (b1484ca)
   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
  46char *get_commit_graph_filename(const char *obj_dir)
  47{
  48        return xstrfmt("%s/info/commit-graph", obj_dir);
  49}
  50
  51static 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
  59struct 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
 180cleanup_fail:
 181        munmap(graph_map, graph_size);
 182        close(fd);
 183        exit(1);
 184}
 185
 186static 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 */
 206static 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 (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
 217            !config_value)
 218                /*
 219                 * This repository is not configured to use commit graphs, so
 220                 * do not load one. (But report commit_graph_attempted anyway
 221                 * so that commit graph loading is not attempted again for this
 222                 * repository.)
 223                 */
 224                return 0;
 225
 226        obj_dir = r->objects->objectdir;
 227        prepare_commit_graph_one(r, obj_dir);
 228        prepare_alt_odb(r);
 229        for (alt = r->objects->alt_odb_list;
 230             !r->objects->commit_graph && alt;
 231             alt = alt->next)
 232                prepare_commit_graph_one(r, alt->path);
 233        return !!r->objects->commit_graph;
 234}
 235
 236static void close_commit_graph(void)
 237{
 238        free_commit_graph(the_repository->objects->commit_graph);
 239        the_repository->objects->commit_graph = NULL;
 240}
 241
 242static int bsearch_graph(struct commit_graph *g, struct object_id *oid, uint32_t *pos)
 243{
 244        return bsearch_hash(oid->hash, g->chunk_oid_fanout,
 245                            g->chunk_oid_lookup, g->hash_len, pos);
 246}
 247
 248static struct commit_list **insert_parent_or_die(struct commit_graph *g,
 249                                                 uint64_t pos,
 250                                                 struct commit_list **pptr)
 251{
 252        struct commit *c;
 253        struct object_id oid;
 254
 255        if (pos >= g->num_commits)
 256                die("invalid parent position %"PRIu64, pos);
 257
 258        hashcpy(oid.hash, g->chunk_oid_lookup + g->hash_len * pos);
 259        c = lookup_commit(the_repository, &oid);
 260        if (!c)
 261                die(_("could not find commit %s"), oid_to_hex(&oid));
 262        c->graph_pos = pos;
 263        return &commit_list_insert(c, pptr)->next;
 264}
 265
 266static void fill_commit_graph_info(struct commit *item, struct commit_graph *g, uint32_t pos)
 267{
 268        const unsigned char *commit_data = g->chunk_commit_data + GRAPH_DATA_WIDTH * pos;
 269        item->graph_pos = pos;
 270        item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 271}
 272
 273static int fill_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t pos)
 274{
 275        uint32_t edge_value;
 276        uint32_t *parent_data_ptr;
 277        uint64_t date_low, date_high;
 278        struct commit_list **pptr;
 279        const unsigned char *commit_data = g->chunk_commit_data + (g->hash_len + 16) * pos;
 280
 281        item->object.parsed = 1;
 282        item->graph_pos = pos;
 283
 284        item->maybe_tree = NULL;
 285
 286        date_high = get_be32(commit_data + g->hash_len + 8) & 0x3;
 287        date_low = get_be32(commit_data + g->hash_len + 12);
 288        item->date = (timestamp_t)((date_high << 32) | date_low);
 289
 290        item->generation = get_be32(commit_data + g->hash_len + 8) >> 2;
 291
 292        pptr = &item->parents;
 293
 294        edge_value = get_be32(commit_data + g->hash_len);
 295        if (edge_value == GRAPH_PARENT_NONE)
 296                return 1;
 297        pptr = insert_parent_or_die(g, edge_value, pptr);
 298
 299        edge_value = get_be32(commit_data + g->hash_len + 4);
 300        if (edge_value == GRAPH_PARENT_NONE)
 301                return 1;
 302        if (!(edge_value & GRAPH_OCTOPUS_EDGES_NEEDED)) {
 303                pptr = insert_parent_or_die(g, edge_value, pptr);
 304                return 1;
 305        }
 306
 307        parent_data_ptr = (uint32_t*)(g->chunk_large_edges +
 308                          4 * (uint64_t)(edge_value & GRAPH_EDGE_LAST_MASK));
 309        do {
 310                edge_value = get_be32(parent_data_ptr);
 311                pptr = insert_parent_or_die(g,
 312                                            edge_value & GRAPH_EDGE_LAST_MASK,
 313                                            pptr);
 314                parent_data_ptr++;
 315        } while (!(edge_value & GRAPH_LAST_EDGE));
 316
 317        return 1;
 318}
 319
 320static int find_commit_in_graph(struct commit *item, struct commit_graph *g, uint32_t *pos)
 321{
 322        if (item->graph_pos != COMMIT_NOT_FROM_GRAPH) {
 323                *pos = item->graph_pos;
 324                return 1;
 325        } else {
 326                return bsearch_graph(g, &(item->object.oid), pos);
 327        }
 328}
 329
 330static int parse_commit_in_graph_one(struct commit_graph *g, struct commit *item)
 331{
 332        uint32_t pos;
 333
 334        if (item->object.parsed)
 335                return 1;
 336
 337        if (find_commit_in_graph(item, g, &pos))
 338                return fill_commit_in_graph(item, g, pos);
 339
 340        return 0;
 341}
 342
 343int parse_commit_in_graph(struct repository *r, struct commit *item)
 344{
 345        if (!prepare_commit_graph(r))
 346                return 0;
 347        return parse_commit_in_graph_one(r->objects->commit_graph, item);
 348}
 349
 350void load_commit_graph_info(struct repository *r, struct commit *item)
 351{
 352        uint32_t pos;
 353        if (!prepare_commit_graph(r))
 354                return;
 355        if (find_commit_in_graph(item, r->objects->commit_graph, &pos))
 356                fill_commit_graph_info(item, r->objects->commit_graph, pos);
 357}
 358
 359static struct tree *load_tree_for_commit(struct commit_graph *g, struct commit *c)
 360{
 361        struct object_id oid;
 362        const unsigned char *commit_data = g->chunk_commit_data +
 363                                           GRAPH_DATA_WIDTH * (c->graph_pos);
 364
 365        hashcpy(oid.hash, commit_data);
 366        c->maybe_tree = lookup_tree(the_repository, &oid);
 367
 368        return c->maybe_tree;
 369}
 370
 371static struct tree *get_commit_tree_in_graph_one(struct commit_graph *g,
 372                                                 const struct commit *c)
 373{
 374        if (c->maybe_tree)
 375                return c->maybe_tree;
 376        if (c->graph_pos == COMMIT_NOT_FROM_GRAPH)
 377                BUG("get_commit_tree_in_graph_one called from non-commit-graph commit");
 378
 379        return load_tree_for_commit(g, (struct commit *)c);
 380}
 381
 382struct tree *get_commit_tree_in_graph(struct repository *r, const struct commit *c)
 383{
 384        return get_commit_tree_in_graph_one(r->objects->commit_graph, c);
 385}
 386
 387static void write_graph_chunk_fanout(struct hashfile *f,
 388                                     struct commit **commits,
 389                                     int nr_commits)
 390{
 391        int i, count = 0;
 392        struct commit **list = commits;
 393
 394        /*
 395         * Write the first-level table (the list is sorted,
 396         * but we use a 256-entry lookup to be able to avoid
 397         * having to do eight extra binary search iterations).
 398         */
 399        for (i = 0; i < 256; i++) {
 400                while (count < nr_commits) {
 401                        if ((*list)->object.oid.hash[0] != i)
 402                                break;
 403                        count++;
 404                        list++;
 405                }
 406
 407                hashwrite_be32(f, count);
 408        }
 409}
 410
 411static void write_graph_chunk_oids(struct hashfile *f, int hash_len,
 412                                   struct commit **commits, int nr_commits)
 413{
 414        struct commit **list = commits;
 415        int count;
 416        for (count = 0; count < nr_commits; count++, list++)
 417                hashwrite(f, (*list)->object.oid.hash, (int)hash_len);
 418}
 419
 420static const unsigned char *commit_to_sha1(size_t index, void *table)
 421{
 422        struct commit **commits = table;
 423        return commits[index]->object.oid.hash;
 424}
 425
 426static void write_graph_chunk_data(struct hashfile *f, int hash_len,
 427                                   struct commit **commits, int nr_commits)
 428{
 429        struct commit **list = commits;
 430        struct commit **last = commits + nr_commits;
 431        uint32_t num_extra_edges = 0;
 432
 433        while (list < last) {
 434                struct commit_list *parent;
 435                int edge_value;
 436                uint32_t packedDate[2];
 437
 438                parse_commit(*list);
 439                hashwrite(f, get_commit_tree_oid(*list)->hash, hash_len);
 440
 441                parent = (*list)->parents;
 442
 443                if (!parent)
 444                        edge_value = GRAPH_PARENT_NONE;
 445                else {
 446                        edge_value = sha1_pos(parent->item->object.oid.hash,
 447                                              commits,
 448                                              nr_commits,
 449                                              commit_to_sha1);
 450
 451                        if (edge_value < 0)
 452                                edge_value = GRAPH_PARENT_MISSING;
 453                }
 454
 455                hashwrite_be32(f, edge_value);
 456
 457                if (parent)
 458                        parent = parent->next;
 459
 460                if (!parent)
 461                        edge_value = GRAPH_PARENT_NONE;
 462                else if (parent->next)
 463                        edge_value = GRAPH_OCTOPUS_EDGES_NEEDED | num_extra_edges;
 464                else {
 465                        edge_value = sha1_pos(parent->item->object.oid.hash,
 466                                              commits,
 467                                              nr_commits,
 468                                              commit_to_sha1);
 469                        if (edge_value < 0)
 470                                edge_value = GRAPH_PARENT_MISSING;
 471                }
 472
 473                hashwrite_be32(f, edge_value);
 474
 475                if (edge_value & GRAPH_OCTOPUS_EDGES_NEEDED) {
 476                        do {
 477                                num_extra_edges++;
 478                                parent = parent->next;
 479                        } while (parent);
 480                }
 481
 482                if (sizeof((*list)->date) > 4)
 483                        packedDate[0] = htonl(((*list)->date >> 32) & 0x3);
 484                else
 485                        packedDate[0] = 0;
 486
 487                packedDate[0] |= htonl((*list)->generation << 2);
 488
 489                packedDate[1] = htonl((*list)->date);
 490                hashwrite(f, packedDate, 8);
 491
 492                list++;
 493        }
 494}
 495
 496static void write_graph_chunk_large_edges(struct hashfile *f,
 497                                          struct commit **commits,
 498                                          int nr_commits)
 499{
 500        struct commit **list = commits;
 501        struct commit **last = commits + nr_commits;
 502        struct commit_list *parent;
 503
 504        while (list < last) {
 505                int num_parents = 0;
 506                for (parent = (*list)->parents; num_parents < 3 && parent;
 507                     parent = parent->next)
 508                        num_parents++;
 509
 510                if (num_parents <= 2) {
 511                        list++;
 512                        continue;
 513                }
 514
 515                /* Since num_parents > 2, this initializer is safe. */
 516                for (parent = (*list)->parents->next; parent; parent = parent->next) {
 517                        int edge_value = sha1_pos(parent->item->object.oid.hash,
 518                                                  commits,
 519                                                  nr_commits,
 520                                                  commit_to_sha1);
 521
 522                        if (edge_value < 0)
 523                                edge_value = GRAPH_PARENT_MISSING;
 524                        else if (!parent->next)
 525                                edge_value |= GRAPH_LAST_EDGE;
 526
 527                        hashwrite_be32(f, edge_value);
 528                }
 529
 530                list++;
 531        }
 532}
 533
 534static int commit_compare(const void *_a, const void *_b)
 535{
 536        const struct object_id *a = (const struct object_id *)_a;
 537        const struct object_id *b = (const struct object_id *)_b;
 538        return oidcmp(a, b);
 539}
 540
 541struct packed_commit_list {
 542        struct commit **list;
 543        int nr;
 544        int alloc;
 545};
 546
 547struct packed_oid_list {
 548        struct object_id *list;
 549        int nr;
 550        int alloc;
 551};
 552
 553static int add_packed_commits(const struct object_id *oid,
 554                              struct packed_git *pack,
 555                              uint32_t pos,
 556                              void *data)
 557{
 558        struct packed_oid_list *list = (struct packed_oid_list*)data;
 559        enum object_type type;
 560        off_t offset = nth_packed_object_offset(pack, pos);
 561        struct object_info oi = OBJECT_INFO_INIT;
 562
 563        oi.typep = &type;
 564        if (packed_object_info(the_repository, pack, offset, &oi) < 0)
 565                die(_("unable to get type of object %s"), oid_to_hex(oid));
 566
 567        if (type != OBJ_COMMIT)
 568                return 0;
 569
 570        ALLOC_GROW(list->list, list->nr + 1, list->alloc);
 571        oidcpy(&(list->list[list->nr]), oid);
 572        list->nr++;
 573
 574        return 0;
 575}
 576
 577static void add_missing_parents(struct packed_oid_list *oids, struct commit *commit)
 578{
 579        struct commit_list *parent;
 580        for (parent = commit->parents; parent; parent = parent->next) {
 581                if (!(parent->item->object.flags & UNINTERESTING)) {
 582                        ALLOC_GROW(oids->list, oids->nr + 1, oids->alloc);
 583                        oidcpy(&oids->list[oids->nr], &(parent->item->object.oid));
 584                        oids->nr++;
 585                        parent->item->object.flags |= UNINTERESTING;
 586                }
 587        }
 588}
 589
 590static void close_reachable(struct packed_oid_list *oids)
 591{
 592        int i;
 593        struct commit *commit;
 594
 595        for (i = 0; i < oids->nr; i++) {
 596                commit = lookup_commit(the_repository, &oids->list[i]);
 597                if (commit)
 598                        commit->object.flags |= UNINTERESTING;
 599        }
 600
 601        /*
 602         * As this loop runs, oids->nr may grow, but not more
 603         * than the number of missing commits in the reachable
 604         * closure.
 605         */
 606        for (i = 0; i < oids->nr; i++) {
 607                commit = lookup_commit(the_repository, &oids->list[i]);
 608
 609                if (commit && !parse_commit(commit))
 610                        add_missing_parents(oids, commit);
 611        }
 612
 613        for (i = 0; i < oids->nr; i++) {
 614                commit = lookup_commit(the_repository, &oids->list[i]);
 615
 616                if (commit)
 617                        commit->object.flags &= ~UNINTERESTING;
 618        }
 619}
 620
 621static void compute_generation_numbers(struct packed_commit_list* commits)
 622{
 623        int i;
 624        struct commit_list *list = NULL;
 625
 626        for (i = 0; i < commits->nr; i++) {
 627                if (commits->list[i]->generation != GENERATION_NUMBER_INFINITY &&
 628                    commits->list[i]->generation != GENERATION_NUMBER_ZERO)
 629                        continue;
 630
 631                commit_list_insert(commits->list[i], &list);
 632                while (list) {
 633                        struct commit *current = list->item;
 634                        struct commit_list *parent;
 635                        int all_parents_computed = 1;
 636                        uint32_t max_generation = 0;
 637
 638                        for (parent = current->parents; parent; parent = parent->next) {
 639                                if (parent->item->generation == GENERATION_NUMBER_INFINITY ||
 640                                    parent->item->generation == GENERATION_NUMBER_ZERO) {
 641                                        all_parents_computed = 0;
 642                                        commit_list_insert(parent->item, &list);
 643                                        break;
 644                                } else if (parent->item->generation > max_generation) {
 645                                        max_generation = parent->item->generation;
 646                                }
 647                        }
 648
 649                        if (all_parents_computed) {
 650                                current->generation = max_generation + 1;
 651                                pop_commit(&list);
 652
 653                                if (current->generation > GENERATION_NUMBER_MAX)
 654                                        current->generation = GENERATION_NUMBER_MAX;
 655                        }
 656                }
 657        }
 658}
 659
 660static int add_ref_to_list(const char *refname,
 661                           const struct object_id *oid,
 662                           int flags, void *cb_data)
 663{
 664        struct string_list *list = (struct string_list *)cb_data;
 665
 666        string_list_append(list, oid_to_hex(oid));
 667        return 0;
 668}
 669
 670void write_commit_graph_reachable(const char *obj_dir, int append)
 671{
 672        struct string_list list;
 673
 674        string_list_init(&list, 1);
 675        for_each_ref(add_ref_to_list, &list);
 676        write_commit_graph(obj_dir, NULL, &list, append);
 677}
 678
 679void write_commit_graph(const char *obj_dir,
 680                        struct string_list *pack_indexes,
 681                        struct string_list *commit_hex,
 682                        int append)
 683{
 684        struct packed_oid_list oids;
 685        struct packed_commit_list commits;
 686        struct hashfile *f;
 687        uint32_t i, count_distinct = 0;
 688        char *graph_name;
 689        struct lock_file lk = LOCK_INIT;
 690        uint32_t chunk_ids[5];
 691        uint64_t chunk_offsets[5];
 692        int num_chunks;
 693        int num_extra_edges;
 694        struct commit_list *parent;
 695
 696        oids.nr = 0;
 697        oids.alloc = approximate_object_count() / 4;
 698
 699        if (append) {
 700                prepare_commit_graph_one(the_repository, obj_dir);
 701                if (the_repository->objects->commit_graph)
 702                        oids.alloc += the_repository->objects->commit_graph->num_commits;
 703        }
 704
 705        if (oids.alloc < 1024)
 706                oids.alloc = 1024;
 707        ALLOC_ARRAY(oids.list, oids.alloc);
 708
 709        if (append && the_repository->objects->commit_graph) {
 710                struct commit_graph *commit_graph =
 711                        the_repository->objects->commit_graph;
 712                for (i = 0; i < commit_graph->num_commits; i++) {
 713                        const unsigned char *hash = commit_graph->chunk_oid_lookup +
 714                                commit_graph->hash_len * i;
 715                        hashcpy(oids.list[oids.nr++].hash, hash);
 716                }
 717        }
 718
 719        if (pack_indexes) {
 720                struct strbuf packname = STRBUF_INIT;
 721                int dirlen;
 722                strbuf_addf(&packname, "%s/pack/", obj_dir);
 723                dirlen = packname.len;
 724                for (i = 0; i < pack_indexes->nr; i++) {
 725                        struct packed_git *p;
 726                        strbuf_setlen(&packname, dirlen);
 727                        strbuf_addstr(&packname, pack_indexes->items[i].string);
 728                        p = add_packed_git(packname.buf, packname.len, 1);
 729                        if (!p)
 730                                die(_("error adding pack %s"), packname.buf);
 731                        if (open_pack_index(p))
 732                                die(_("error opening index for %s"), packname.buf);
 733                        for_each_object_in_pack(p, add_packed_commits, &oids, 0);
 734                        close_pack(p);
 735                }
 736                strbuf_release(&packname);
 737        }
 738
 739        if (commit_hex) {
 740                for (i = 0; i < commit_hex->nr; i++) {
 741                        const char *end;
 742                        struct object_id oid;
 743                        struct commit *result;
 744
 745                        if (commit_hex->items[i].string &&
 746                            parse_oid_hex(commit_hex->items[i].string, &oid, &end))
 747                                continue;
 748
 749                        result = lookup_commit_reference_gently(the_repository, &oid, 1);
 750
 751                        if (result) {
 752                                ALLOC_GROW(oids.list, oids.nr + 1, oids.alloc);
 753                                oidcpy(&oids.list[oids.nr], &(result->object.oid));
 754                                oids.nr++;
 755                        }
 756                }
 757        }
 758
 759        if (!pack_indexes && !commit_hex)
 760                for_each_packed_object(add_packed_commits, &oids, 0);
 761
 762        close_reachable(&oids);
 763
 764        QSORT(oids.list, oids.nr, commit_compare);
 765
 766        count_distinct = 1;
 767        for (i = 1; i < oids.nr; i++) {
 768                if (oidcmp(&oids.list[i-1], &oids.list[i]))
 769                        count_distinct++;
 770        }
 771
 772        if (count_distinct >= GRAPH_PARENT_MISSING)
 773                die(_("the commit graph format cannot write %d commits"), count_distinct);
 774
 775        commits.nr = 0;
 776        commits.alloc = count_distinct;
 777        ALLOC_ARRAY(commits.list, commits.alloc);
 778
 779        num_extra_edges = 0;
 780        for (i = 0; i < oids.nr; i++) {
 781                int num_parents = 0;
 782                if (i > 0 && !oidcmp(&oids.list[i-1], &oids.list[i]))
 783                        continue;
 784
 785                commits.list[commits.nr] = lookup_commit(the_repository, &oids.list[i]);
 786                parse_commit(commits.list[commits.nr]);
 787
 788                for (parent = commits.list[commits.nr]->parents;
 789                     parent; parent = parent->next)
 790                        num_parents++;
 791
 792                if (num_parents > 2)
 793                        num_extra_edges += num_parents - 1;
 794
 795                commits.nr++;
 796        }
 797        num_chunks = num_extra_edges ? 4 : 3;
 798
 799        if (commits.nr >= GRAPH_PARENT_MISSING)
 800                die(_("too many commits to write graph"));
 801
 802        compute_generation_numbers(&commits);
 803
 804        graph_name = get_commit_graph_filename(obj_dir);
 805        if (safe_create_leading_directories(graph_name))
 806                die_errno(_("unable to create leading directories of %s"),
 807                          graph_name);
 808
 809        hold_lock_file_for_update(&lk, graph_name, LOCK_DIE_ON_ERROR);
 810        f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 811
 812        hashwrite_be32(f, GRAPH_SIGNATURE);
 813
 814        hashwrite_u8(f, GRAPH_VERSION);
 815        hashwrite_u8(f, GRAPH_OID_VERSION);
 816        hashwrite_u8(f, num_chunks);
 817        hashwrite_u8(f, 0); /* unused padding byte */
 818
 819        chunk_ids[0] = GRAPH_CHUNKID_OIDFANOUT;
 820        chunk_ids[1] = GRAPH_CHUNKID_OIDLOOKUP;
 821        chunk_ids[2] = GRAPH_CHUNKID_DATA;
 822        if (num_extra_edges)
 823                chunk_ids[3] = GRAPH_CHUNKID_LARGEEDGES;
 824        else
 825                chunk_ids[3] = 0;
 826        chunk_ids[4] = 0;
 827
 828        chunk_offsets[0] = 8 + (num_chunks + 1) * GRAPH_CHUNKLOOKUP_WIDTH;
 829        chunk_offsets[1] = chunk_offsets[0] + GRAPH_FANOUT_SIZE;
 830        chunk_offsets[2] = chunk_offsets[1] + GRAPH_OID_LEN * commits.nr;
 831        chunk_offsets[3] = chunk_offsets[2] + (GRAPH_OID_LEN + 16) * commits.nr;
 832        chunk_offsets[4] = chunk_offsets[3] + 4 * num_extra_edges;
 833
 834        for (i = 0; i <= num_chunks; i++) {
 835                uint32_t chunk_write[3];
 836
 837                chunk_write[0] = htonl(chunk_ids[i]);
 838                chunk_write[1] = htonl(chunk_offsets[i] >> 32);
 839                chunk_write[2] = htonl(chunk_offsets[i] & 0xffffffff);
 840                hashwrite(f, chunk_write, 12);
 841        }
 842
 843        write_graph_chunk_fanout(f, commits.list, commits.nr);
 844        write_graph_chunk_oids(f, GRAPH_OID_LEN, commits.list, commits.nr);
 845        write_graph_chunk_data(f, GRAPH_OID_LEN, commits.list, commits.nr);
 846        write_graph_chunk_large_edges(f, commits.list, commits.nr);
 847
 848        close_commit_graph();
 849        finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC);
 850        commit_lock_file(&lk);
 851
 852        free(oids.list);
 853        oids.alloc = 0;
 854        oids.nr = 0;
 855}
 856
 857#define VERIFY_COMMIT_GRAPH_ERROR_HASH 2
 858static int verify_commit_graph_error;
 859
 860static void graph_report(const char *fmt, ...)
 861{
 862        va_list ap;
 863
 864        verify_commit_graph_error = 1;
 865        va_start(ap, fmt);
 866        vfprintf(stderr, fmt, ap);
 867        fprintf(stderr, "\n");
 868        va_end(ap);
 869}
 870
 871#define GENERATION_ZERO_EXISTS 1
 872#define GENERATION_NUMBER_EXISTS 2
 873
 874int verify_commit_graph(struct repository *r, struct commit_graph *g)
 875{
 876        uint32_t i, cur_fanout_pos = 0;
 877        struct object_id prev_oid, cur_oid, checksum;
 878        int generation_zero = 0;
 879        struct hashfile *f;
 880        int devnull;
 881
 882        if (!g) {
 883                graph_report("no commit-graph file loaded");
 884                return 1;
 885        }
 886
 887        verify_commit_graph_error = 0;
 888
 889        if (!g->chunk_oid_fanout)
 890                graph_report("commit-graph is missing the OID Fanout chunk");
 891        if (!g->chunk_oid_lookup)
 892                graph_report("commit-graph is missing the OID Lookup chunk");
 893        if (!g->chunk_commit_data)
 894                graph_report("commit-graph is missing the Commit Data chunk");
 895
 896        if (verify_commit_graph_error)
 897                return verify_commit_graph_error;
 898
 899        devnull = open("/dev/null", O_WRONLY);
 900        f = hashfd(devnull, NULL);
 901        hashwrite(f, g->data, g->data_len - g->hash_len);
 902        finalize_hashfile(f, checksum.hash, CSUM_CLOSE);
 903        if (hashcmp(checksum.hash, g->data + g->data_len - g->hash_len)) {
 904                graph_report(_("the commit-graph file has incorrect checksum and is likely corrupt"));
 905                verify_commit_graph_error = VERIFY_COMMIT_GRAPH_ERROR_HASH;
 906        }
 907
 908        for (i = 0; i < g->num_commits; i++) {
 909                struct commit *graph_commit;
 910
 911                hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 912
 913                if (i && oidcmp(&prev_oid, &cur_oid) >= 0)
 914                        graph_report("commit-graph has incorrect OID order: %s then %s",
 915                                     oid_to_hex(&prev_oid),
 916                                     oid_to_hex(&cur_oid));
 917
 918                oidcpy(&prev_oid, &cur_oid);
 919
 920                while (cur_oid.hash[0] > cur_fanout_pos) {
 921                        uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 922
 923                        if (i != fanout_value)
 924                                graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 925                                             cur_fanout_pos, fanout_value, i);
 926                        cur_fanout_pos++;
 927                }
 928
 929                graph_commit = lookup_commit(r, &cur_oid);
 930                if (!parse_commit_in_graph_one(g, graph_commit))
 931                        graph_report("failed to parse %s from commit-graph",
 932                                     oid_to_hex(&cur_oid));
 933        }
 934
 935        while (cur_fanout_pos < 256) {
 936                uint32_t fanout_value = get_be32(g->chunk_oid_fanout + cur_fanout_pos);
 937
 938                if (g->num_commits != fanout_value)
 939                        graph_report("commit-graph has incorrect fanout value: fanout[%d] = %u != %u",
 940                                     cur_fanout_pos, fanout_value, i);
 941
 942                cur_fanout_pos++;
 943        }
 944
 945        if (verify_commit_graph_error & ~VERIFY_COMMIT_GRAPH_ERROR_HASH)
 946                return verify_commit_graph_error;
 947
 948        for (i = 0; i < g->num_commits; i++) {
 949                struct commit *graph_commit, *odb_commit;
 950                struct commit_list *graph_parents, *odb_parents;
 951                uint32_t max_generation = 0;
 952
 953                hashcpy(cur_oid.hash, g->chunk_oid_lookup + g->hash_len * i);
 954
 955                graph_commit = lookup_commit(r, &cur_oid);
 956                odb_commit = (struct commit *)create_object(r, cur_oid.hash, alloc_commit_node(r));
 957                if (parse_commit_internal(odb_commit, 0, 0)) {
 958                        graph_report("failed to parse %s from object database",
 959                                     oid_to_hex(&cur_oid));
 960                        continue;
 961                }
 962
 963                if (oidcmp(&get_commit_tree_in_graph_one(g, graph_commit)->object.oid,
 964                           get_commit_tree_oid(odb_commit)))
 965                        graph_report("root tree OID for commit %s in commit-graph is %s != %s",
 966                                     oid_to_hex(&cur_oid),
 967                                     oid_to_hex(get_commit_tree_oid(graph_commit)),
 968                                     oid_to_hex(get_commit_tree_oid(odb_commit)));
 969
 970                graph_parents = graph_commit->parents;
 971                odb_parents = odb_commit->parents;
 972
 973                while (graph_parents) {
 974                        if (odb_parents == NULL) {
 975                                graph_report("commit-graph parent list for commit %s is too long",
 976                                             oid_to_hex(&cur_oid));
 977                                break;
 978                        }
 979
 980                        if (oidcmp(&graph_parents->item->object.oid, &odb_parents->item->object.oid))
 981                                graph_report("commit-graph parent for %s is %s != %s",
 982                                             oid_to_hex(&cur_oid),
 983                                             oid_to_hex(&graph_parents->item->object.oid),
 984                                             oid_to_hex(&odb_parents->item->object.oid));
 985
 986                        if (graph_parents->item->generation > max_generation)
 987                                max_generation = graph_parents->item->generation;
 988
 989                        graph_parents = graph_parents->next;
 990                        odb_parents = odb_parents->next;
 991                }
 992
 993                if (odb_parents != NULL)
 994                        graph_report("commit-graph parent list for commit %s terminates early",
 995                                     oid_to_hex(&cur_oid));
 996
 997                if (!graph_commit->generation) {
 998                        if (generation_zero == GENERATION_NUMBER_EXISTS)
 999                                graph_report("commit-graph has generation number zero for commit %s, but non-zero elsewhere",
1000                                             oid_to_hex(&cur_oid));
1001                        generation_zero = GENERATION_ZERO_EXISTS;
1002                } else if (generation_zero == GENERATION_ZERO_EXISTS)
1003                        graph_report("commit-graph has non-zero generation number for commit %s, but zero elsewhere",
1004                                     oid_to_hex(&cur_oid));
1005
1006                if (generation_zero == GENERATION_ZERO_EXISTS)
1007                        continue;
1008
1009                /*
1010                 * If one of our parents has generation GENERATION_NUMBER_MAX, then
1011                 * our generation is also GENERATION_NUMBER_MAX. Decrement to avoid
1012                 * extra logic in the following condition.
1013                 */
1014                if (max_generation == GENERATION_NUMBER_MAX)
1015                        max_generation--;
1016
1017                if (graph_commit->generation != max_generation + 1)
1018                        graph_report("commit-graph generation for commit %s is %u != %u",
1019                                     oid_to_hex(&cur_oid),
1020                                     graph_commit->generation,
1021                                     max_generation + 1);
1022
1023                if (graph_commit->date != odb_commit->date)
1024                        graph_report("commit date for commit %s in commit-graph is %"PRItime" != %"PRItime,
1025                                     oid_to_hex(&cur_oid),
1026                                     graph_commit->date,
1027                                     odb_commit->date);
1028        }
1029
1030        return verify_commit_graph_error;
1031}
1032
1033void free_commit_graph(struct commit_graph *g)
1034{
1035        if (!g)
1036                return;
1037        if (g->graph_fd >= 0) {
1038                munmap((void *)g->data, g->data_len);
1039                g->data = NULL;
1040                close(g->graph_fd);
1041        }
1042        free(g);
1043}