git.oblomov.eu Git - git/blob - midx.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "csum-file.h"
   4 #include "dir.h"
   5 #include "lockfile.h"
   6 #include "packfile.h"
   7 #include "object-store.h"
   8 #include "sha1-lookup.h"
   9 #include "midx.h"
  10 #include "progress.h"
  11
  12 #define MIDX_SIGNATURE 0x4d494458 /* "MIDX" */
  13 #define MIDX_VERSION 1
  14 #define MIDX_BYTE_FILE_VERSION 4
  15 #define MIDX_BYTE_HASH_VERSION 5
  16 #define MIDX_BYTE_NUM_CHUNKS 6
  17 #define MIDX_BYTE_NUM_PACKS 8
  18 #define MIDX_HASH_VERSION 1
  19 #define MIDX_HEADER_SIZE 12
  20 #define MIDX_HASH_LEN 20
  21 #define MIDX_MIN_SIZE (MIDX_HEADER_SIZE + MIDX_HASH_LEN)
  22
  23 #define MIDX_MAX_CHUNKS 5
  24 #define MIDX_CHUNK_ALIGNMENT 4
  25 #define MIDX_CHUNKID_PACKNAMES 0x504e414d /* "PNAM" */
  26 #define MIDX_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  27 #define MIDX_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  28 #define MIDX_CHUNKID_OBJECTOFFSETS 0x4f4f4646 /* "OOFF" */
  29 #define MIDX_CHUNKID_LARGEOFFSETS 0x4c4f4646 /* "LOFF" */
  30 #define MIDX_CHUNKLOOKUP_WIDTH (sizeof(uint32_t) + sizeof(uint64_t))
  31 #define MIDX_CHUNK_FANOUT_SIZE (sizeof(uint32_t) * 256)
  32 #define MIDX_CHUNK_OFFSET_WIDTH (2 * sizeof(uint32_t))
  33 #define MIDX_CHUNK_LARGE_OFFSET_WIDTH (sizeof(uint64_t))
  34 #define MIDX_LARGE_OFFSET_NEEDED 0x80000000
  35
  36 static char *get_midx_filename(const char *object_dir)
  37 {
  38         return xstrfmt("%s/pack/multi-pack-index", object_dir);
  39 }
  40
  41 struct multi_pack_index *load_multi_pack_index(const char *object_dir, int local)
  42 {
  43         struct multi_pack_index *m = NULL;
  44         int fd;
  45         struct stat st;
  46         size_t midx_size;
  47         void *midx_map = NULL;
  48         uint32_t hash_version;
  49         char *midx_name = get_midx_filename(object_dir);
  50         uint32_t i;
  51         const char *cur_pack_name;
  52
  53         fd = git_open(midx_name);
  54
  55         if (fd < 0)
  56                 goto cleanup_fail;
  57         if (fstat(fd, &st)) {
  58                 error_errno(_("failed to read %s"), midx_name);
  59                 goto cleanup_fail;
  60         }
  61
  62         midx_size = xsize_t(st.st_size);
  63
  64         if (midx_size < MIDX_MIN_SIZE) {
  65                 error(_("multi-pack-index file %s is too small"), midx_name);
  66                 goto cleanup_fail;
  67         }
  68
  69         FREE_AND_NULL(midx_name);
  70
  71         midx_map = xmmap(NULL, midx_size, PROT_READ, MAP_PRIVATE, fd, 0);
  72
  73         FLEX_ALLOC_MEM(m, object_dir, object_dir, strlen(object_dir));
  74         m->fd = fd;
  75         m->data = midx_map;
  76         m->data_len = midx_size;
  77         m->local = local;
  78
  79         m->signature = get_be32(m->data);
  80         if (m->signature != MIDX_SIGNATURE)
  81                 die(_("multi-pack-index signature 0x%08x does not match signature 0x%08x"),
  82                       m->signature, MIDX_SIGNATURE);
  83
  84         m->version = m->data[MIDX_BYTE_FILE_VERSION];
  85         if (m->version != MIDX_VERSION)
  86                 die(_("multi-pack-index version %d not recognized"),
  87                       m->version);
  88
  89         hash_version = m->data[MIDX_BYTE_HASH_VERSION];
  90         if (hash_version != MIDX_HASH_VERSION)
  91                 die(_("hash version %u does not match"), hash_version);
  92         m->hash_len = MIDX_HASH_LEN;
  93
  94         m->num_chunks = m->data[MIDX_BYTE_NUM_CHUNKS];
  95
  96         m->num_packs = get_be32(m->data + MIDX_BYTE_NUM_PACKS);
  97
  98         for (i = 0; i < m->num_chunks; i++) {
  99                 uint32_t chunk_id = get_be32(m->data + MIDX_HEADER_SIZE +
 100                                              MIDX_CHUNKLOOKUP_WIDTH * i);
 101                 uint64_t chunk_offset = get_be64(m->data + MIDX_HEADER_SIZE + 4 +
 102                                                  MIDX_CHUNKLOOKUP_WIDTH * i);
 103
 104                 if (chunk_offset >= m->data_len)
 105                         die(_("invalid chunk offset (too large)"));
 106
 107                 switch (chunk_id) {
 108                         case MIDX_CHUNKID_PACKNAMES:
 109                                 m->chunk_pack_names = m->data + chunk_offset;
 110                                 break;
 111
 112                         case MIDX_CHUNKID_OIDFANOUT:
 113                                 m->chunk_oid_fanout = (uint32_t *)(m->data + chunk_offset);
 114                                 break;
 115
 116                         case MIDX_CHUNKID_OIDLOOKUP:
 117                                 m->chunk_oid_lookup = m->data + chunk_offset;
 118                                 break;
 119
 120                         case MIDX_CHUNKID_OBJECTOFFSETS:
 121                                 m->chunk_object_offsets = m->data + chunk_offset;
 122                                 break;
 123
 124                         case MIDX_CHUNKID_LARGEOFFSETS:
 125                                 m->chunk_large_offsets = m->data + chunk_offset;
 126                                 break;
 127
 128                         case 0:
 129                                 die(_("terminating multi-pack-index chunk id appears earlier than expected"));
 130                                 break;
 131
 132                         default:
 133                                 /*
 134                                  * Do nothing on unrecognized chunks, allowing future
 135                                  * extensions to add optional chunks.
 136                                  */
 137                                 break;
 138                 }
 139         }
 140
 141         if (!m->chunk_pack_names)
 142                 die(_("multi-pack-index missing required pack-name chunk"));
 143         if (!m->chunk_oid_fanout)
 144                 die(_("multi-pack-index missing required OID fanout chunk"));
 145         if (!m->chunk_oid_lookup)
 146                 die(_("multi-pack-index missing required OID lookup chunk"));
 147         if (!m->chunk_object_offsets)
 148                 die(_("multi-pack-index missing required object offsets chunk"));
 149
 150         m->num_objects = ntohl(m->chunk_oid_fanout[255]);
 151
 152         m->pack_names = xcalloc(m->num_packs, sizeof(*m->pack_names));
 153         m->packs = xcalloc(m->num_packs, sizeof(*m->packs));
 154
 155         cur_pack_name = (const char *)m->chunk_pack_names;
 156         for (i = 0; i < m->num_packs; i++) {
 157                 m->pack_names[i] = cur_pack_name;
 158
 159                 cur_pack_name += strlen(cur_pack_name) + 1;
 160
 161                 if (i && strcmp(m->pack_names[i], m->pack_names[i - 1]) <= 0)
 162                         die(_("multi-pack-index pack names out of order: '%s' before '%s'"),
 163                               m->pack_names[i - 1],
 164                               m->pack_names[i]);
 165         }
 166
 167         return m;
 168
 169 cleanup_fail:
 170         free(m);
 171         free(midx_name);
 172         if (midx_map)
 173                 munmap(midx_map, midx_size);
 174         if (0 <= fd)
 175                 close(fd);
 176         return NULL;
 177 }
 178
 179 void close_midx(struct multi_pack_index *m)
 180 {
 181         uint32_t i;
 182
 183         if (!m)
 184                 return;
 185
 186         munmap((unsigned char *)m->data, m->data_len);
 187         close(m->fd);
 188         m->fd = -1;
 189
 190         for (i = 0; i < m->num_packs; i++) {
 191                 if (m->packs[i]) {
 192                         close_pack(m->packs[i]);
 193                         free(m->packs[i]);
 194                 }
 195         }
 196         FREE_AND_NULL(m->packs);
 197         FREE_AND_NULL(m->pack_names);
 198 }
 199
 200 int prepare_midx_pack(struct multi_pack_index *m, uint32_t pack_int_id)
 201 {
 202         struct strbuf pack_name = STRBUF_INIT;
 203
 204         if (pack_int_id >= m->num_packs)
 205                 die(_("bad pack-int-id: %u (%u total packs)"),
 206                     pack_int_id, m->num_packs);
 207
 208         if (m->packs[pack_int_id])
 209                 return 0;
 210
 211         strbuf_addf(&pack_name, "%s/pack/%s", m->object_dir,
 212                     m->pack_names[pack_int_id]);
 213
 214         m->packs[pack_int_id] = add_packed_git(pack_name.buf, pack_name.len, m->local);
 215         strbuf_release(&pack_name);
 216         return !m->packs[pack_int_id];
 217 }
 218
 219 int bsearch_midx(const struct object_id *oid, struct multi_pack_index *m, uint32_t *result)
 220 {
 221         return bsearch_hash(oid->hash, m->chunk_oid_fanout, m->chunk_oid_lookup,
 222                             MIDX_HASH_LEN, result);
 223 }
 224
 225 struct object_id *nth_midxed_object_oid(struct object_id *oid,
 226                                         struct multi_pack_index *m,
 227                                         uint32_t n)
 228 {
 229         if (n >= m->num_objects)
 230                 return NULL;
 231
 232         hashcpy(oid->hash, m->chunk_oid_lookup + m->hash_len * n);
 233         return oid;
 234 }
 235
 236 static off_t nth_midxed_offset(struct multi_pack_index *m, uint32_t pos)
 237 {
 238         const unsigned char *offset_data;
 239         uint32_t offset32;
 240
 241         offset_data = m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH;
 242         offset32 = get_be32(offset_data + sizeof(uint32_t));
 243
 244         if (m->chunk_large_offsets && offset32 & MIDX_LARGE_OFFSET_NEEDED) {
 245                 if (sizeof(off_t) < sizeof(uint64_t))
 246                         die(_("multi-pack-index stores a 64-bit offset, but off_t is too small"));
 247
 248                 offset32 ^= MIDX_LARGE_OFFSET_NEEDED;
 249                 return get_be64(m->chunk_large_offsets + sizeof(uint64_t) * offset32);
 250         }
 251
 252         return offset32;
 253 }
 254
 255 static uint32_t nth_midxed_pack_int_id(struct multi_pack_index *m, uint32_t pos)
 256 {
 257         return get_be32(m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH);
 258 }
 259
 260 static int nth_midxed_pack_entry(struct multi_pack_index *m, struct pack_entry *e, uint32_t pos)
 261 {
 262         uint32_t pack_int_id;
 263         struct packed_git *p;
 264
 265         if (pos >= m->num_objects)
 266                 return 0;
 267
 268         pack_int_id = nth_midxed_pack_int_id(m, pos);
 269
 270         if (prepare_midx_pack(m, pack_int_id))
 271                 die(_("error preparing packfile from multi-pack-index"));
 272         p = m->packs[pack_int_id];
 273
 274         /*
 275         * We are about to tell the caller where they can locate the
 276         * requested object.  We better make sure the packfile is
 277         * still here and can be accessed before supplying that
 278         * answer, as it may have been deleted since the MIDX was
 279         * loaded!
 280         */
 281         if (!is_pack_valid(p))
 282                 return 0;
 283
 284         if (p->num_bad_objects) {
 285                 uint32_t i;
 286                 struct object_id oid;
 287                 nth_midxed_object_oid(&oid, m, pos);
 288                 for (i = 0; i < p->num_bad_objects; i++)
 289                         if (hasheq(oid.hash,
 290                                    p->bad_object_sha1 + the_hash_algo->rawsz * i))
 291                                 return 0;
 292         }
 293
 294         e->offset = nth_midxed_offset(m, pos);
 295         e->p = p;
 296
 297         return 1;
 298 }
 299
 300 int fill_midx_entry(const struct object_id *oid, struct pack_entry *e, struct multi_pack_index *m)
 301 {
 302         uint32_t pos;
 303
 304         if (!bsearch_midx(oid, m, &pos))
 305                 return 0;
 306
 307         return nth_midxed_pack_entry(m, e, pos);
 308 }
 309
 310 /* Match "foo.idx" against either "foo.pack" _or_ "foo.idx". */
 311 static int cmp_idx_or_pack_name(const char *idx_or_pack_name,
 312                                 const char *idx_name)
 313 {
 314         /* Skip past any initial matching prefix. */
 315         while (*idx_name && *idx_name == *idx_or_pack_name) {
 316                 idx_name++;
 317                 idx_or_pack_name++;
 318         }
 319
 320         /*
 321          * If we didn't match completely, we may have matched "pack-1234." and
 322          * be left with "idx" and "pack" respectively, which is also OK. We do
 323          * not have to check for "idx" and "idx", because that would have been
 324          * a complete match (and in that case these strcmps will be false, but
 325          * we'll correctly return 0 from the final strcmp() below.
 326          *
 327          * Technically this matches "fooidx" and "foopack", but we'd never have
 328          * such names in the first place.
 329          */
 330         if (!strcmp(idx_name, "idx") && !strcmp(idx_or_pack_name, "pack"))
 331                 return 0;
 332
 333         /*
 334          * This not only checks for a complete match, but also orders based on
 335          * the first non-identical character, which means our ordering will
 336          * match a raw strcmp(). That makes it OK to use this to binary search
 337          * a naively-sorted list.
 338          */
 339         return strcmp(idx_or_pack_name, idx_name);
 340 }
 341
 342 int midx_contains_pack(struct multi_pack_index *m, const char *idx_or_pack_name)
 343 {
 344         uint32_t first = 0, last = m->num_packs;
 345
 346         while (first < last) {
 347                 uint32_t mid = first + (last - first) / 2;
 348                 const char *current;
 349                 int cmp;
 350
 351                 current = m->pack_names[mid];
 352                 cmp = cmp_idx_or_pack_name(idx_or_pack_name, current);
 353                 if (!cmp)
 354                         return 1;
 355                 if (cmp > 0) {
 356                         first = mid + 1;
 357                         continue;
 358                 }
 359                 last = mid;
 360         }
 361
 362         return 0;
 363 }
 364
 365 int prepare_multi_pack_index_one(struct repository *r, const char *object_dir, int local)
 366 {
 367         struct multi_pack_index *m;
 368         struct multi_pack_index *m_search;
 369         int config_value;
 370         static int env_value = -1;
 371
 372         if (env_value < 0)
 373                 env_value = git_env_bool(GIT_TEST_MULTI_PACK_INDEX, 0);
 374
 375         if (!env_value &&
 376             (repo_config_get_bool(r, "core.multipackindex", &config_value) ||
 377             !config_value))
 378                 return 0;
 379
 380         for (m_search = r->objects->multi_pack_index; m_search; m_search = m_search->next)
 381                 if (!strcmp(object_dir, m_search->object_dir))
 382                         return 1;
 383
 384         m = load_multi_pack_index(object_dir, local);
 385
 386         if (m) {
 387                 m->next = r->objects->multi_pack_index;
 388                 r->objects->multi_pack_index = m;
 389                 return 1;
 390         }
 391
 392         return 0;
 393 }
 394
 395 static size_t write_midx_header(struct hashfile *f,
 396                                 unsigned char num_chunks,
 397                                 uint32_t num_packs)
 398 {
 399         unsigned char byte_values[4];
 400
 401         hashwrite_be32(f, MIDX_SIGNATURE);
 402         byte_values[0] = MIDX_VERSION;
 403         byte_values[1] = MIDX_HASH_VERSION;
 404         byte_values[2] = num_chunks;
 405         byte_values[3] = 0; /* unused */
 406         hashwrite(f, byte_values, sizeof(byte_values));
 407         hashwrite_be32(f, num_packs);
 408
 409         return MIDX_HEADER_SIZE;
 410 }
 411
 412 struct pack_list {
 413         struct packed_git **list;
 414         char **names;
 415         uint32_t nr;
 416         uint32_t alloc_list;
 417         uint32_t alloc_names;
 418         size_t pack_name_concat_len;
 419         struct multi_pack_index *m;
 420 };
 421
 422 static void add_pack_to_midx(const char *full_path, size_t full_path_len,
 423                              const char *file_name, void *data)
 424 {
 425         struct pack_list *packs = (struct pack_list *)data;
 426
 427         if (ends_with(file_name, ".idx")) {
 428                 if (packs->m && midx_contains_pack(packs->m, file_name))
 429                         return;
 430
 431                 ALLOC_GROW(packs->list, packs->nr + 1, packs->alloc_list);
 432                 ALLOC_GROW(packs->names, packs->nr + 1, packs->alloc_names);
 433
 434                 packs->list[packs->nr] = add_packed_git(full_path,
 435                                                         full_path_len,
 436                                                         0);
 437
 438                 if (!packs->list[packs->nr]) {
 439                         warning(_("failed to add packfile '%s'"),
 440                                 full_path);
 441                         return;
 442                 }
 443
 444                 if (open_pack_index(packs->list[packs->nr])) {
 445                         warning(_("failed to open pack-index '%s'"),
 446                                 full_path);
 447                         close_pack(packs->list[packs->nr]);
 448                         FREE_AND_NULL(packs->list[packs->nr]);
 449                         return;
 450                 }
 451
 452                 packs->names[packs->nr] = xstrdup(file_name);
 453                 packs->pack_name_concat_len += strlen(file_name) + 1;
 454                 packs->nr++;
 455         }
 456 }
 457
 458 struct pack_pair {
 459         uint32_t pack_int_id;
 460         char *pack_name;
 461 };
 462
 463 static int pack_pair_compare(const void *_a, const void *_b)
 464 {
 465         struct pack_pair *a = (struct pack_pair *)_a;
 466         struct pack_pair *b = (struct pack_pair *)_b;
 467         return strcmp(a->pack_name, b->pack_name);
 468 }
 469
 470 static void sort_packs_by_name(char **pack_names, uint32_t nr_packs, uint32_t *perm)
 471 {
 472         uint32_t i;
 473         struct pack_pair *pairs;
 474
 475         ALLOC_ARRAY(pairs, nr_packs);
 476
 477         for (i = 0; i < nr_packs; i++) {
 478                 pairs[i].pack_int_id = i;
 479                 pairs[i].pack_name = pack_names[i];
 480         }
 481
 482         QSORT(pairs, nr_packs, pack_pair_compare);
 483
 484         for (i = 0; i < nr_packs; i++) {
 485                 pack_names[i] = pairs[i].pack_name;
 486                 perm[pairs[i].pack_int_id] = i;
 487         }
 488
 489         free(pairs);
 490 }
 491
 492 struct pack_midx_entry {
 493         struct object_id oid;
 494         uint32_t pack_int_id;
 495         time_t pack_mtime;
 496         uint64_t offset;
 497 };
 498
 499 static int midx_oid_compare(const void *_a, const void *_b)
 500 {
 501         const struct pack_midx_entry *a = (const struct pack_midx_entry *)_a;
 502         const struct pack_midx_entry *b = (const struct pack_midx_entry *)_b;
 503         int cmp = oidcmp(&a->oid, &b->oid);
 504
 505         if (cmp)
 506                 return cmp;
 507
 508         if (a->pack_mtime > b->pack_mtime)
 509                 return -1;
 510         else if (a->pack_mtime < b->pack_mtime)
 511                 return 1;
 512
 513         return a->pack_int_id - b->pack_int_id;
 514 }
 515
 516 static int nth_midxed_pack_midx_entry(struct multi_pack_index *m,
 517                                       uint32_t *pack_perm,
 518                                       struct pack_midx_entry *e,
 519                                       uint32_t pos)
 520 {
 521         if (pos >= m->num_objects)
 522                 return 1;
 523
 524         nth_midxed_object_oid(&e->oid, m, pos);
 525         e->pack_int_id = pack_perm[nth_midxed_pack_int_id(m, pos)];
 526         e->offset = nth_midxed_offset(m, pos);
 527
 528         /* consider objects in midx to be from "old" packs */
 529         e->pack_mtime = 0;
 530         return 0;
 531 }
 532
 533 static void fill_pack_entry(uint32_t pack_int_id,
 534                             struct packed_git *p,
 535                             uint32_t cur_object,
 536                             struct pack_midx_entry *entry)
 537 {
 538         if (!nth_packed_object_oid(&entry->oid, p, cur_object))
 539                 die(_("failed to locate object %d in packfile"), cur_object);
 540
 541         entry->pack_int_id = pack_int_id;
 542         entry->pack_mtime = p->mtime;
 543
 544         entry->offset = nth_packed_object_offset(p, cur_object);
 545 }
 546
 547 /*
 548  * It is possible to artificially get into a state where there are many
 549  * duplicate copies of objects. That can create high memory pressure if
 550  * we are to create a list of all objects before de-duplication. To reduce
 551  * this memory pressure without a significant performance drop, automatically
 552  * group objects by the first byte of their object id. Use the IDX fanout
 553  * tables to group the data, copy to a local array, then sort.
 554  *
 555  * Copy only the de-duplicated entries (selected by most-recent modified time
 556  * of a packfile containing the object).
 557  */
 558 static struct pack_midx_entry *get_sorted_entries(struct multi_pack_index *m,
 559                                                   struct packed_git **p,
 560                                                   uint32_t *perm,
 561                                                   uint32_t nr_packs,
 562                                                   uint32_t *nr_objects)
 563 {
 564         uint32_t cur_fanout, cur_pack, cur_object;
 565         uint32_t alloc_fanout, alloc_objects, total_objects = 0;
 566         struct pack_midx_entry *entries_by_fanout = NULL;
 567         struct pack_midx_entry *deduplicated_entries = NULL;
 568         uint32_t start_pack = m ? m->num_packs : 0;
 569
 570         for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++)
 571                 total_objects += p[cur_pack]->num_objects;
 572
 573         /*
 574          * As we de-duplicate by fanout value, we expect the fanout
 575          * slices to be evenly distributed, with some noise. Hence,
 576          * allocate slightly more than one 256th.
 577          */
 578         alloc_objects = alloc_fanout = total_objects > 3200 ? total_objects / 200 : 16;
 579
 580         ALLOC_ARRAY(entries_by_fanout, alloc_fanout);
 581         ALLOC_ARRAY(deduplicated_entries, alloc_objects);
 582         *nr_objects = 0;
 583
 584         for (cur_fanout = 0; cur_fanout < 256; cur_fanout++) {
 585                 uint32_t nr_fanout = 0;
 586
 587                 if (m) {
 588                         uint32_t start = 0, end;
 589
 590                         if (cur_fanout)
 591                                 start = ntohl(m->chunk_oid_fanout[cur_fanout - 1]);
 592                         end = ntohl(m->chunk_oid_fanout[cur_fanout]);
 593
 594                         for (cur_object = start; cur_object < end; cur_object++) {
 595                                 ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
 596                                 nth_midxed_pack_midx_entry(m, perm,
 597                                                            &entries_by_fanout[nr_fanout],
 598                                                            cur_object);
 599                                 nr_fanout++;
 600                         }
 601                 }
 602
 603                 for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++) {
 604                         uint32_t start = 0, end;
 605
 606                         if (cur_fanout)
 607                                 start = get_pack_fanout(p[cur_pack], cur_fanout - 1);
 608                         end = get_pack_fanout(p[cur_pack], cur_fanout);
 609
 610                         for (cur_object = start; cur_object < end; cur_object++) {
 611                                 ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
 612                                 fill_pack_entry(perm[cur_pack], p[cur_pack], cur_object, &entries_by_fanout[nr_fanout]);
 613                                 nr_fanout++;
 614                         }
 615                 }
 616
 617                 QSORT(entries_by_fanout, nr_fanout, midx_oid_compare);
 618
 619                 /*
 620                  * The batch is now sorted by OID and then mtime (descending).
 621                  * Take only the first duplicate.
 622                  */
 623                 for (cur_object = 0; cur_object < nr_fanout; cur_object++) {
 624                         if (cur_object && oideq(&entries_by_fanout[cur_object - 1].oid,
 625                                                 &entries_by_fanout[cur_object].oid))
 626                                 continue;
 627
 628                         ALLOC_GROW(deduplicated_entries, *nr_objects + 1, alloc_objects);
 629                         memcpy(&deduplicated_entries[*nr_objects],
 630                                &entries_by_fanout[cur_object],
 631                                sizeof(struct pack_midx_entry));
 632                         (*nr_objects)++;
 633                 }
 634         }
 635
 636         free(entries_by_fanout);
 637         return deduplicated_entries;
 638 }
 639
 640 static size_t write_midx_pack_names(struct hashfile *f,
 641                                     char **pack_names,
 642                                     uint32_t num_packs)
 643 {
 644         uint32_t i;
 645         unsigned char padding[MIDX_CHUNK_ALIGNMENT];
 646         size_t written = 0;
 647
 648         for (i = 0; i < num_packs; i++) {
 649                 size_t writelen = strlen(pack_names[i]) + 1;
 650
 651                 if (i && strcmp(pack_names[i], pack_names[i - 1]) <= 0)
 652                         BUG("incorrect pack-file order: %s before %s",
 653                             pack_names[i - 1],
 654                             pack_names[i]);
 655
 656                 hashwrite(f, pack_names[i], writelen);
 657                 written += writelen;
 658         }
 659
 660         /* add padding to be aligned */
 661         i = MIDX_CHUNK_ALIGNMENT - (written % MIDX_CHUNK_ALIGNMENT);
 662         if (i < MIDX_CHUNK_ALIGNMENT) {
 663                 memset(padding, 0, sizeof(padding));
 664                 hashwrite(f, padding, i);
 665                 written += i;
 666         }
 667
 668         return written;
 669 }
 670
 671 static size_t write_midx_oid_fanout(struct hashfile *f,
 672                                     struct pack_midx_entry *objects,
 673                                     uint32_t nr_objects)
 674 {
 675         struct pack_midx_entry *list = objects;
 676         struct pack_midx_entry *last = objects + nr_objects;
 677         uint32_t count = 0;
 678         uint32_t i;
 679
 680         /*
 681         * Write the first-level table (the list is sorted,
 682         * but we use a 256-entry lookup to be able to avoid
 683         * having to do eight extra binary search iterations).
 684         */
 685         for (i = 0; i < 256; i++) {
 686                 struct pack_midx_entry *next = list;
 687
 688                 while (next < last && next->oid.hash[0] == i) {
 689                         count++;
 690                         next++;
 691                 }
 692
 693                 hashwrite_be32(f, count);
 694                 list = next;
 695         }
 696
 697         return MIDX_CHUNK_FANOUT_SIZE;
 698 }
 699
 700 static size_t write_midx_oid_lookup(struct hashfile *f, unsigned char hash_len,
 701                                     struct pack_midx_entry *objects,
 702                                     uint32_t nr_objects)
 703 {
 704         struct pack_midx_entry *list = objects;
 705         uint32_t i;
 706         size_t written = 0;
 707
 708         for (i = 0; i < nr_objects; i++) {
 709                 struct pack_midx_entry *obj = list++;
 710
 711                 if (i < nr_objects - 1) {
 712                         struct pack_midx_entry *next = list;
 713                         if (oidcmp(&obj->oid, &next->oid) >= 0)
 714                                 BUG("OIDs not in order: %s >= %s",
 715                                     oid_to_hex(&obj->oid),
 716                                     oid_to_hex(&next->oid));
 717                 }
 718
 719                 hashwrite(f, obj->oid.hash, (int)hash_len);
 720                 written += hash_len;
 721         }
 722
 723         return written;
 724 }
 725
 726 static size_t write_midx_object_offsets(struct hashfile *f, int large_offset_needed,
 727                                         struct pack_midx_entry *objects, uint32_t nr_objects)
 728 {
 729         struct pack_midx_entry *list = objects;
 730         uint32_t i, nr_large_offset = 0;
 731         size_t written = 0;
 732
 733         for (i = 0; i < nr_objects; i++) {
 734                 struct pack_midx_entry *obj = list++;
 735
 736                 hashwrite_be32(f, obj->pack_int_id);
 737
 738                 if (large_offset_needed && obj->offset >> 31)
 739                         hashwrite_be32(f, MIDX_LARGE_OFFSET_NEEDED | nr_large_offset++);
 740                 else if (!large_offset_needed && obj->offset >> 32)
 741                         BUG("object %s requires a large offset (%"PRIx64") but the MIDX is not writing large offsets!",
 742                             oid_to_hex(&obj->oid),
 743                             obj->offset);
 744                 else
 745                         hashwrite_be32(f, (uint32_t)obj->offset);
 746
 747                 written += MIDX_CHUNK_OFFSET_WIDTH;
 748         }
 749
 750         return written;
 751 }
 752
 753 static size_t write_midx_large_offsets(struct hashfile *f, uint32_t nr_large_offset,
 754                                        struct pack_midx_entry *objects, uint32_t nr_objects)
 755 {
 756         struct pack_midx_entry *list = objects, *end = objects + nr_objects;
 757         size_t written = 0;
 758
 759         while (nr_large_offset) {
 760                 struct pack_midx_entry *obj;
 761                 uint64_t offset;
 762
 763                 if (list >= end)
 764                         BUG("too many large-offset objects");
 765
 766                 obj = list++;
 767                 offset = obj->offset;
 768
 769                 if (!(offset >> 31))
 770                         continue;
 771
 772                 hashwrite_be32(f, offset >> 32);
 773                 hashwrite_be32(f, offset & 0xffffffffUL);
 774                 written += 2 * sizeof(uint32_t);
 775
 776                 nr_large_offset--;
 777         }
 778
 779         return written;
 780 }
 781
 782 int write_midx_file(const char *object_dir)
 783 {
 784         unsigned char cur_chunk, num_chunks = 0;
 785         char *midx_name;
 786         uint32_t i;
 787         struct hashfile *f = NULL;
 788         struct lock_file lk;
 789         struct pack_list packs;
 790         uint32_t *pack_perm = NULL;
 791         uint64_t written = 0;
 792         uint32_t chunk_ids[MIDX_MAX_CHUNKS + 1];
 793         uint64_t chunk_offsets[MIDX_MAX_CHUNKS + 1];
 794         uint32_t nr_entries, num_large_offsets = 0;
 795         struct pack_midx_entry *entries = NULL;
 796         int large_offsets_needed = 0;
 797
 798         midx_name = get_midx_filename(object_dir);
 799         if (safe_create_leading_directories(midx_name)) {
 800                 UNLEAK(midx_name);
 801                 die_errno(_("unable to create leading directories of %s"),
 802                           midx_name);
 803         }
 804
 805         packs.m = load_multi_pack_index(object_dir, 1);
 806
 807         packs.nr = 0;
 808         packs.alloc_list = packs.m ? packs.m->num_packs : 16;
 809         packs.alloc_names = packs.alloc_list;
 810         packs.list = NULL;
 811         packs.names = NULL;
 812         packs.pack_name_concat_len = 0;
 813         ALLOC_ARRAY(packs.list, packs.alloc_list);
 814         ALLOC_ARRAY(packs.names, packs.alloc_names);
 815
 816         if (packs.m) {
 817                 for (i = 0; i < packs.m->num_packs; i++) {
 818                         ALLOC_GROW(packs.list, packs.nr + 1, packs.alloc_list);
 819                         ALLOC_GROW(packs.names, packs.nr + 1, packs.alloc_names);
 820
 821                         packs.list[packs.nr] = NULL;
 822                         packs.names[packs.nr] = xstrdup(packs.m->pack_names[i]);
 823                         packs.pack_name_concat_len += strlen(packs.names[packs.nr]) + 1;
 824                         packs.nr++;
 825                 }
 826         }
 827
 828         for_each_file_in_pack_dir(object_dir, add_pack_to_midx, &packs);
 829
 830         if (packs.m && packs.nr == packs.m->num_packs)
 831                 goto cleanup;
 832
 833         if (packs.pack_name_concat_len % MIDX_CHUNK_ALIGNMENT)
 834                 packs.pack_name_concat_len += MIDX_CHUNK_ALIGNMENT -
 835                                               (packs.pack_name_concat_len % MIDX_CHUNK_ALIGNMENT);
 836
 837         ALLOC_ARRAY(pack_perm, packs.nr);
 838         sort_packs_by_name(packs.names, packs.nr, pack_perm);
 839
 840         entries = get_sorted_entries(packs.m, packs.list, pack_perm, packs.nr, &nr_entries);
 841
 842         for (i = 0; i < nr_entries; i++) {
 843                 if (entries[i].offset > 0x7fffffff)
 844                         num_large_offsets++;
 845                 if (entries[i].offset > 0xffffffff)
 846                         large_offsets_needed = 1;
 847         }
 848
 849         hold_lock_file_for_update(&lk, midx_name, LOCK_DIE_ON_ERROR);
 850         f = hashfd(lk.tempfile->fd, lk.tempfile->filename.buf);
 851         FREE_AND_NULL(midx_name);
 852
 853         if (packs.m)
 854                 close_midx(packs.m);
 855
 856         cur_chunk = 0;
 857         num_chunks = large_offsets_needed ? 5 : 4;
 858
 859         written = write_midx_header(f, num_chunks, packs.nr);
 860
 861         chunk_ids[cur_chunk] = MIDX_CHUNKID_PACKNAMES;
 862         chunk_offsets[cur_chunk] = written + (num_chunks + 1) * MIDX_CHUNKLOOKUP_WIDTH;
 863
 864         cur_chunk++;
 865         chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDFANOUT;
 866         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + packs.pack_name_concat_len;
 867
 868         cur_chunk++;
 869         chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDLOOKUP;
 870         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + MIDX_CHUNK_FANOUT_SIZE;
 871
 872         cur_chunk++;
 873         chunk_ids[cur_chunk] = MIDX_CHUNKID_OBJECTOFFSETS;
 874         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * MIDX_HASH_LEN;
 875
 876         cur_chunk++;
 877         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * MIDX_CHUNK_OFFSET_WIDTH;
 878         if (large_offsets_needed) {
 879                 chunk_ids[cur_chunk] = MIDX_CHUNKID_LARGEOFFSETS;
 880
 881                 cur_chunk++;
 882                 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] +
 883                                            num_large_offsets * MIDX_CHUNK_LARGE_OFFSET_WIDTH;
 884         }
 885
 886         chunk_ids[cur_chunk] = 0;
 887
 888         for (i = 0; i <= num_chunks; i++) {
 889                 if (i && chunk_offsets[i] < chunk_offsets[i - 1])
 890                         BUG("incorrect chunk offsets: %"PRIu64" before %"PRIu64,
 891                             chunk_offsets[i - 1],
 892                             chunk_offsets[i]);
 893
 894                 if (chunk_offsets[i] % MIDX_CHUNK_ALIGNMENT)
 895                         BUG("chunk offset %"PRIu64" is not properly aligned",
 896                             chunk_offsets[i]);
 897
 898                 hashwrite_be32(f, chunk_ids[i]);
 899                 hashwrite_be32(f, chunk_offsets[i] >> 32);
 900                 hashwrite_be32(f, chunk_offsets[i]);
 901
 902                 written += MIDX_CHUNKLOOKUP_WIDTH;
 903         }
 904
 905         for (i = 0; i < num_chunks; i++) {
 906                 if (written != chunk_offsets[i])
 907                         BUG("incorrect chunk offset (%"PRIu64" != %"PRIu64") for chunk id %"PRIx32,
 908                             chunk_offsets[i],
 909                             written,
 910                             chunk_ids[i]);
 911
 912                 switch (chunk_ids[i]) {
 913                         case MIDX_CHUNKID_PACKNAMES:
 914                                 written += write_midx_pack_names(f, packs.names, packs.nr);
 915                                 break;
 916
 917                         case MIDX_CHUNKID_OIDFANOUT:
 918                                 written += write_midx_oid_fanout(f, entries, nr_entries);
 919                                 break;
 920
 921                         case MIDX_CHUNKID_OIDLOOKUP:
 922                                 written += write_midx_oid_lookup(f, MIDX_HASH_LEN, entries, nr_entries);
 923                                 break;
 924
 925                         case MIDX_CHUNKID_OBJECTOFFSETS:
 926                                 written += write_midx_object_offsets(f, large_offsets_needed, entries, nr_entries);
 927                                 break;
 928
 929                         case MIDX_CHUNKID_LARGEOFFSETS:
 930                                 written += write_midx_large_offsets(f, num_large_offsets, entries, nr_entries);
 931                                 break;
 932
 933                         default:
 934                                 BUG("trying to write unknown chunk id %"PRIx32,
 935                                     chunk_ids[i]);
 936                 }
 937         }
 938
 939         if (written != chunk_offsets[num_chunks])
 940                 BUG("incorrect final offset %"PRIu64" != %"PRIu64,
 941                     written,
 942                     chunk_offsets[num_chunks]);
 943
 944         finalize_hashfile(f, NULL, CSUM_FSYNC | CSUM_HASH_IN_STREAM);
 945         commit_lock_file(&lk);
 946
 947 cleanup:
 948         for (i = 0; i < packs.nr; i++) {
 949                 if (packs.list[i]) {
 950                         close_pack(packs.list[i]);
 951                         free(packs.list[i]);
 952                 }
 953                 free(packs.names[i]);
 954         }
 955
 956         free(packs.list);
 957         free(packs.names);
 958         free(entries);
 959         free(pack_perm);
 960         free(midx_name);
 961         return 0;
 962 }
 963
 964 void clear_midx_file(struct repository *r)
 965 {
 966         char *midx = get_midx_filename(r->objects->odb->path);
 967
 968         if (r->objects && r->objects->multi_pack_index) {
 969                 close_midx(r->objects->multi_pack_index);
 970                 r->objects->multi_pack_index = NULL;
 971         }
 972
 973         if (remove_path(midx)) {
 974                 UNLEAK(midx);
 975                 die(_("failed to clear multi-pack-index at %s"), midx);
 976         }
 977
 978         free(midx);
 979 }
 980
 981 static int verify_midx_error;
 982
 983 static void midx_report(const char *fmt, ...)
 984 {
 985         va_list ap;
 986         verify_midx_error = 1;
 987         va_start(ap, fmt);
 988         vfprintf(stderr, fmt, ap);
 989         fprintf(stderr, "\n");
 990         va_end(ap);
 991 }
 992
 993 int verify_midx_file(const char *object_dir)
 994 {
 995         uint32_t i;
 996         struct progress *progress;
 997         struct multi_pack_index *m = load_multi_pack_index(object_dir, 1);
 998         verify_midx_error = 0;
 999
1000         if (!m)
1001                 return 0;
1002
1003         for (i = 0; i < m->num_packs; i++) {
1004                 if (prepare_midx_pack(m, i))
1005                         midx_report("failed to load pack in position %d", i);
1006         }
1007
1008         for (i = 0; i < 255; i++) {
1009                 uint32_t oid_fanout1 = ntohl(m->chunk_oid_fanout[i]);
1010                 uint32_t oid_fanout2 = ntohl(m->chunk_oid_fanout[i + 1]);
1011
1012                 if (oid_fanout1 > oid_fanout2)
1013                         midx_report(_("oid fanout out of order: fanout[%d] = %"PRIx32" > %"PRIx32" = fanout[%d]"),
1014                                     i, oid_fanout1, oid_fanout2, i + 1);
1015         }
1016
1017         for (i = 0; i < m->num_objects - 1; i++) {
1018                 struct object_id oid1, oid2;
1019
1020                 nth_midxed_object_oid(&oid1, m, i);
1021                 nth_midxed_object_oid(&oid2, m, i + 1);
1022
1023                 if (oidcmp(&oid1, &oid2) >= 0)
1024                         midx_report(_("oid lookup out of order: oid[%d] = %s >= %s = oid[%d]"),
1025                                     i, oid_to_hex(&oid1), oid_to_hex(&oid2), i + 1);
1026         }
1027
1028         progress = start_progress(_("Verifying object offsets"), m->num_objects);
1029         for (i = 0; i < m->num_objects; i++) {
1030                 struct object_id oid;
1031                 struct pack_entry e;
1032                 off_t m_offset, p_offset;
1033
1034                 nth_midxed_object_oid(&oid, m, i);
1035                 if (!fill_midx_entry(&oid, &e, m)) {
1036                         midx_report(_("failed to load pack entry for oid[%d] = %s"),
1037                                     i, oid_to_hex(&oid));
1038                         continue;
1039                 }
1040
1041                 if (open_pack_index(e.p)) {
1042                         midx_report(_("failed to load pack-index for packfile %s"),
1043                                     e.p->pack_name);
1044                         break;
1045                 }
1046
1047                 m_offset = e.offset;
1048                 p_offset = find_pack_entry_one(oid.hash, e.p);
1049
1050                 if (m_offset != p_offset)
1051                         midx_report(_("incorrect object offset for oid[%d] = %s: %"PRIx64" != %"PRIx64),
1052                                     i, oid_to_hex(&oid), m_offset, p_offset);
1053
1054                 display_progress(progress, i + 1);
1055         }
1056         stop_progress(&progress);
1057
1058         return verify_midx_error;
1059 }