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