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 "hash-lookup.h"
   9 #include "midx.h"
  10 #include "progress.h"
  11 #include "trace2.h"
  12 #include "run-command.h"
  13 #include "repository.h"
  14
  15 #define MIDX_SIGNATURE 0x4d494458 /* "MIDX" */
  16 #define MIDX_VERSION 1
  17 #define MIDX_BYTE_FILE_VERSION 4
  18 #define MIDX_BYTE_HASH_VERSION 5
  19 #define MIDX_BYTE_NUM_CHUNKS 6
  20 #define MIDX_BYTE_NUM_PACKS 8
  21 #define MIDX_HEADER_SIZE 12
  22 #define MIDX_MIN_SIZE (MIDX_HEADER_SIZE + the_hash_algo->rawsz)
  23
  24 #define MIDX_MAX_CHUNKS 5
  25 #define MIDX_CHUNK_ALIGNMENT 4
  26 #define MIDX_CHUNKID_PACKNAMES 0x504e414d /* "PNAM" */
  27 #define MIDX_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
  28 #define MIDX_CHUNKID_OIDLOOKUP 0x4f49444c /* "OIDL" */
  29 #define MIDX_CHUNKID_OBJECTOFFSETS 0x4f4f4646 /* "OOFF" */
  30 #define MIDX_CHUNKID_LARGEOFFSETS 0x4c4f4646 /* "LOFF" */
  31 #define MIDX_CHUNKLOOKUP_WIDTH (sizeof(uint32_t) + sizeof(uint64_t))
  32 #define MIDX_CHUNK_FANOUT_SIZE (sizeof(uint32_t) * 256)
  33 #define MIDX_CHUNK_OFFSET_WIDTH (2 * sizeof(uint32_t))
  34 #define MIDX_CHUNK_LARGE_OFFSET_WIDTH (sizeof(uint64_t))
  35 #define MIDX_LARGE_OFFSET_NEEDED 0x80000000
  36
  37 #define PACK_EXPIRED UINT_MAX
  38
  39 static uint8_t oid_version(void)
  40 {
  41         switch (hash_algo_by_ptr(the_hash_algo)) {
  42         case GIT_HASH_SHA1:
  43                 return 1;
  44         case GIT_HASH_SHA256:
  45                 return 2;
  46         default:
  47                 die(_("invalid hash version"));
  48         }
  49 }
  50
  51 static char *get_midx_filename(const char *object_dir)
  52 {
  53         return xstrfmt("%s/pack/multi-pack-index", object_dir);
  54 }
  55
  56 struct multi_pack_index *load_multi_pack_index(const char *object_dir, int local)
  57 {
  58         struct multi_pack_index *m = NULL;
  59         int fd;
  60         struct stat st;
  61         size_t midx_size;
  62         void *midx_map = NULL;
  63         uint32_t hash_version;
  64         char *midx_name = get_midx_filename(object_dir);
  65         uint32_t i;
  66         const char *cur_pack_name;
  67
  68         fd = git_open(midx_name);
  69
  70         if (fd < 0)
  71                 goto cleanup_fail;
  72         if (fstat(fd, &st)) {
  73                 error_errno(_("failed to read %s"), midx_name);
  74                 goto cleanup_fail;
  75         }
  76
  77         midx_size = xsize_t(st.st_size);
  78
  79         if (midx_size < MIDX_MIN_SIZE) {
  80                 error(_("multi-pack-index file %s is too small"), midx_name);
  81                 goto cleanup_fail;
  82         }
  83
  84         FREE_AND_NULL(midx_name);
  85
  86         midx_map = xmmap(NULL, midx_size, PROT_READ, MAP_PRIVATE, fd, 0);
  87         close(fd);
  88
  89         FLEX_ALLOC_STR(m, object_dir, object_dir);
  90         m->data = midx_map;
  91         m->data_len = midx_size;
  92         m->local = local;
  93
  94         m->signature = get_be32(m->data);
  95         if (m->signature != MIDX_SIGNATURE)
  96                 die(_("multi-pack-index signature 0x%08x does not match signature 0x%08x"),
  97                       m->signature, MIDX_SIGNATURE);
  98
  99         m->version = m->data[MIDX_BYTE_FILE_VERSION];
 100         if (m->version != MIDX_VERSION)
 101                 die(_("multi-pack-index version %d not recognized"),
 102                       m->version);
 103
 104         hash_version = m->data[MIDX_BYTE_HASH_VERSION];
 105         if (hash_version != oid_version()) {
 106                 error(_("multi-pack-index hash version %u does not match version %u"),
 107                       hash_version, oid_version());
 108                 goto cleanup_fail;
 109         }
 110         m->hash_len = the_hash_algo->rawsz;
 111
 112         m->num_chunks = m->data[MIDX_BYTE_NUM_CHUNKS];
 113
 114         m->num_packs = get_be32(m->data + MIDX_BYTE_NUM_PACKS);
 115
 116         for (i = 0; i < m->num_chunks; i++) {
 117                 uint32_t chunk_id = get_be32(m->data + MIDX_HEADER_SIZE +
 118                                              MIDX_CHUNKLOOKUP_WIDTH * i);
 119                 uint64_t chunk_offset = get_be64(m->data + MIDX_HEADER_SIZE + 4 +
 120                                                  MIDX_CHUNKLOOKUP_WIDTH * i);
 121
 122                 if (chunk_offset >= m->data_len)
 123                         die(_("invalid chunk offset (too large)"));
 124
 125                 switch (chunk_id) {
 126                         case MIDX_CHUNKID_PACKNAMES:
 127                                 m->chunk_pack_names = m->data + chunk_offset;
 128                                 break;
 129
 130                         case MIDX_CHUNKID_OIDFANOUT:
 131                                 m->chunk_oid_fanout = (uint32_t *)(m->data + chunk_offset);
 132                                 break;
 133
 134                         case MIDX_CHUNKID_OIDLOOKUP:
 135                                 m->chunk_oid_lookup = m->data + chunk_offset;
 136                                 break;
 137
 138                         case MIDX_CHUNKID_OBJECTOFFSETS:
 139                                 m->chunk_object_offsets = m->data + chunk_offset;
 140                                 break;
 141
 142                         case MIDX_CHUNKID_LARGEOFFSETS:
 143                                 m->chunk_large_offsets = m->data + chunk_offset;
 144                                 break;
 145
 146                         case 0:
 147                                 die(_("terminating multi-pack-index chunk id appears earlier than expected"));
 148                                 break;
 149
 150                         default:
 151                                 /*
 152                                  * Do nothing on unrecognized chunks, allowing future
 153                                  * extensions to add optional chunks.
 154                                  */
 155                                 break;
 156                 }
 157         }
 158
 159         if (!m->chunk_pack_names)
 160                 die(_("multi-pack-index missing required pack-name chunk"));
 161         if (!m->chunk_oid_fanout)
 162                 die(_("multi-pack-index missing required OID fanout chunk"));
 163         if (!m->chunk_oid_lookup)
 164                 die(_("multi-pack-index missing required OID lookup chunk"));
 165         if (!m->chunk_object_offsets)
 166                 die(_("multi-pack-index missing required object offsets chunk"));
 167
 168         m->num_objects = ntohl(m->chunk_oid_fanout[255]);
 169
 170         m->pack_names = xcalloc(m->num_packs, sizeof(*m->pack_names));
 171         m->packs = xcalloc(m->num_packs, sizeof(*m->packs));
 172
 173         cur_pack_name = (const char *)m->chunk_pack_names;
 174         for (i = 0; i < m->num_packs; i++) {
 175                 m->pack_names[i] = cur_pack_name;
 176
 177                 cur_pack_name += strlen(cur_pack_name) + 1;
 178
 179                 if (i && strcmp(m->pack_names[i], m->pack_names[i - 1]) <= 0)
 180                         die(_("multi-pack-index pack names out of order: '%s' before '%s'"),
 181                               m->pack_names[i - 1],
 182                               m->pack_names[i]);
 183         }
 184
 185         trace2_data_intmax("midx", the_repository, "load/num_packs", m->num_packs);
 186         trace2_data_intmax("midx", the_repository, "load/num_objects", m->num_objects);
 187
 188         return m;
 189
 190 cleanup_fail:
 191         free(m);
 192         free(midx_name);
 193         if (midx_map)
 194                 munmap(midx_map, midx_size);
 195         if (0 <= fd)
 196                 close(fd);
 197         return NULL;
 198 }
 199
 200 void close_midx(struct multi_pack_index *m)
 201 {
 202         uint32_t i;
 203
 204         if (!m)
 205                 return;
 206
 207         munmap((unsigned char *)m->data, m->data_len);
 208
 209         for (i = 0; i < m->num_packs; i++) {
 210                 if (m->packs[i])
 211                         m->packs[i]->multi_pack_index = 0;
 212         }
 213         FREE_AND_NULL(m->packs);
 214         FREE_AND_NULL(m->pack_names);
 215 }
 216
 217 int prepare_midx_pack(struct repository *r, struct multi_pack_index *m, uint32_t pack_int_id)
 218 {
 219         struct strbuf pack_name = STRBUF_INIT;
 220         struct packed_git *p;
 221
 222         if (pack_int_id >= m->num_packs)
 223                 die(_("bad pack-int-id: %u (%u total packs)"),
 224                     pack_int_id, m->num_packs);
 225
 226         if (m->packs[pack_int_id])
 227                 return 0;
 228
 229         strbuf_addf(&pack_name, "%s/pack/%s", m->object_dir,
 230                     m->pack_names[pack_int_id]);
 231
 232         p = add_packed_git(pack_name.buf, pack_name.len, m->local);
 233         strbuf_release(&pack_name);
 234
 235         if (!p)
 236                 return 1;
 237
 238         p->multi_pack_index = 1;
 239         m->packs[pack_int_id] = p;
 240         install_packed_git(r, p);
 241         list_add_tail(&p->mru, &r->objects->packed_git_mru);
 242
 243         return 0;
 244 }
 245
 246 int bsearch_midx(const struct object_id *oid, struct multi_pack_index *m, uint32_t *result)
 247 {
 248         return bsearch_hash(oid->hash, m->chunk_oid_fanout, m->chunk_oid_lookup,
 249                             the_hash_algo->rawsz, result);
 250 }
 251
 252 struct object_id *nth_midxed_object_oid(struct object_id *oid,
 253                                         struct multi_pack_index *m,
 254                                         uint32_t n)
 255 {
 256         if (n >= m->num_objects)
 257                 return NULL;
 258
 259         hashcpy(oid->hash, m->chunk_oid_lookup + m->hash_len * n);
 260         return oid;
 261 }
 262
 263 static off_t nth_midxed_offset(struct multi_pack_index *m, uint32_t pos)
 264 {
 265         const unsigned char *offset_data;
 266         uint32_t offset32;
 267
 268         offset_data = m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH;
 269         offset32 = get_be32(offset_data + sizeof(uint32_t));
 270
 271         if (m->chunk_large_offsets && offset32 & MIDX_LARGE_OFFSET_NEEDED) {
 272                 if (sizeof(off_t) < sizeof(uint64_t))
 273                         die(_("multi-pack-index stores a 64-bit offset, but off_t is too small"));
 274
 275                 offset32 ^= MIDX_LARGE_OFFSET_NEEDED;
 276                 return get_be64(m->chunk_large_offsets + sizeof(uint64_t) * offset32);
 277         }
 278
 279         return offset32;
 280 }
 281
 282 static uint32_t nth_midxed_pack_int_id(struct multi_pack_index *m, uint32_t pos)
 283 {
 284         return get_be32(m->chunk_object_offsets + pos * MIDX_CHUNK_OFFSET_WIDTH);
 285 }
 286
 287 static int nth_midxed_pack_entry(struct repository *r,
 288                                  struct multi_pack_index *m,
 289                                  struct pack_entry *e,
 290                                  uint32_t pos)
 291 {
 292         uint32_t pack_int_id;
 293         struct packed_git *p;
 294
 295         if (pos >= m->num_objects)
 296                 return 0;
 297
 298         pack_int_id = nth_midxed_pack_int_id(m, pos);
 299
 300         if (prepare_midx_pack(r, m, pack_int_id))
 301                 return 0;
 302         p = m->packs[pack_int_id];
 303
 304         /*
 305         * We are about to tell the caller where they can locate the
 306         * requested object.  We better make sure the packfile is
 307         * still here and can be accessed before supplying that
 308         * answer, as it may have been deleted since the MIDX was
 309         * loaded!
 310         */
 311         if (!is_pack_valid(p))
 312                 return 0;
 313
 314         if (p->num_bad_objects) {
 315                 uint32_t i;
 316                 struct object_id oid;
 317                 nth_midxed_object_oid(&oid, m, pos);
 318                 for (i = 0; i < p->num_bad_objects; i++)
 319                         if (hasheq(oid.hash,
 320                                    p->bad_object_sha1 + the_hash_algo->rawsz * i))
 321                                 return 0;
 322         }
 323
 324         e->offset = nth_midxed_offset(m, pos);
 325         e->p = p;
 326
 327         return 1;
 328 }
 329
 330 int fill_midx_entry(struct repository * r,
 331                     const struct object_id *oid,
 332                     struct pack_entry *e,
 333                     struct multi_pack_index *m)
 334 {
 335         uint32_t pos;
 336
 337         if (!bsearch_midx(oid, m, &pos))
 338                 return 0;
 339
 340         return nth_midxed_pack_entry(r, m, e, pos);
 341 }
 342
 343 /* Match "foo.idx" against either "foo.pack" _or_ "foo.idx". */
 344 static int cmp_idx_or_pack_name(const char *idx_or_pack_name,
 345                                 const char *idx_name)
 346 {
 347         /* Skip past any initial matching prefix. */
 348         while (*idx_name && *idx_name == *idx_or_pack_name) {
 349                 idx_name++;
 350                 idx_or_pack_name++;
 351         }
 352
 353         /*
 354          * If we didn't match completely, we may have matched "pack-1234." and
 355          * be left with "idx" and "pack" respectively, which is also OK. We do
 356          * not have to check for "idx" and "idx", because that would have been
 357          * a complete match (and in that case these strcmps will be false, but
 358          * we'll correctly return 0 from the final strcmp() below.
 359          *
 360          * Technically this matches "fooidx" and "foopack", but we'd never have
 361          * such names in the first place.
 362          */
 363         if (!strcmp(idx_name, "idx") && !strcmp(idx_or_pack_name, "pack"))
 364                 return 0;
 365
 366         /*
 367          * This not only checks for a complete match, but also orders based on
 368          * the first non-identical character, which means our ordering will
 369          * match a raw strcmp(). That makes it OK to use this to binary search
 370          * a naively-sorted list.
 371          */
 372         return strcmp(idx_or_pack_name, idx_name);
 373 }
 374
 375 int midx_contains_pack(struct multi_pack_index *m, const char *idx_or_pack_name)
 376 {
 377         uint32_t first = 0, last = m->num_packs;
 378
 379         while (first < last) {
 380                 uint32_t mid = first + (last - first) / 2;
 381                 const char *current;
 382                 int cmp;
 383
 384                 current = m->pack_names[mid];
 385                 cmp = cmp_idx_or_pack_name(idx_or_pack_name, current);
 386                 if (!cmp)
 387                         return 1;
 388                 if (cmp > 0) {
 389                         first = mid + 1;
 390                         continue;
 391                 }
 392                 last = mid;
 393         }
 394
 395         return 0;
 396 }
 397
 398 int prepare_multi_pack_index_one(struct repository *r, const char *object_dir, int local)
 399 {
 400         struct multi_pack_index *m;
 401         struct multi_pack_index *m_search;
 402
 403         prepare_repo_settings(r);
 404         if (!r->settings.core_multi_pack_index)
 405                 return 0;
 406
 407         for (m_search = r->objects->multi_pack_index; m_search; m_search = m_search->next)
 408                 if (!strcmp(object_dir, m_search->object_dir))
 409                         return 1;
 410
 411         m = load_multi_pack_index(object_dir, local);
 412
 413         if (m) {
 414                 struct multi_pack_index *mp = r->objects->multi_pack_index;
 415                 if (mp) {
 416                         m->next = mp->next;
 417                         mp->next = m;
 418                 } else
 419                         r->objects->multi_pack_index = m;
 420                 return 1;
 421         }
 422
 423         return 0;
 424 }
 425
 426 static size_t write_midx_header(struct hashfile *f,
 427                                 unsigned char num_chunks,
 428                                 uint32_t num_packs)
 429 {
 430         hashwrite_be32(f, MIDX_SIGNATURE);
 431         hashwrite_u8(f, MIDX_VERSION);
 432         hashwrite_u8(f, oid_version());
 433         hashwrite_u8(f, num_chunks);
 434         hashwrite_u8(f, 0); /* unused */
 435         hashwrite_be32(f, num_packs);
 436
 437         return MIDX_HEADER_SIZE;
 438 }
 439
 440 struct pack_info {
 441         uint32_t orig_pack_int_id;
 442         char *pack_name;
 443         struct packed_git *p;
 444         unsigned expired : 1;
 445 };
 446
 447 static int pack_info_compare(const void *_a, const void *_b)
 448 {
 449         struct pack_info *a = (struct pack_info *)_a;
 450         struct pack_info *b = (struct pack_info *)_b;
 451         return strcmp(a->pack_name, b->pack_name);
 452 }
 453
 454 struct pack_list {
 455         struct pack_info *info;
 456         uint32_t nr;
 457         uint32_t alloc;
 458         struct multi_pack_index *m;
 459         struct progress *progress;
 460         unsigned pack_paths_checked;
 461 };
 462
 463 static void add_pack_to_midx(const char *full_path, size_t full_path_len,
 464                              const char *file_name, void *data)
 465 {
 466         struct pack_list *packs = (struct pack_list *)data;
 467
 468         if (ends_with(file_name, ".idx")) {
 469                 display_progress(packs->progress, ++packs->pack_paths_checked);
 470                 if (packs->m && midx_contains_pack(packs->m, file_name))
 471                         return;
 472
 473                 ALLOC_GROW(packs->info, packs->nr + 1, packs->alloc);
 474
 475                 packs->info[packs->nr].p = add_packed_git(full_path,
 476                                                           full_path_len,
 477                                                           0);
 478
 479                 if (!packs->info[packs->nr].p) {
 480                         warning(_("failed to add packfile '%s'"),
 481                                 full_path);
 482                         return;
 483                 }
 484
 485                 if (open_pack_index(packs->info[packs->nr].p)) {
 486                         warning(_("failed to open pack-index '%s'"),
 487                                 full_path);
 488                         close_pack(packs->info[packs->nr].p);
 489                         FREE_AND_NULL(packs->info[packs->nr].p);
 490                         return;
 491                 }
 492
 493                 packs->info[packs->nr].pack_name = xstrdup(file_name);
 494                 packs->info[packs->nr].orig_pack_int_id = packs->nr;
 495                 packs->info[packs->nr].expired = 0;
 496                 packs->nr++;
 497         }
 498 }
 499
 500 struct pack_midx_entry {
 501         struct object_id oid;
 502         uint32_t pack_int_id;
 503         time_t pack_mtime;
 504         uint64_t offset;
 505 };
 506
 507 static int midx_oid_compare(const void *_a, const void *_b)
 508 {
 509         const struct pack_midx_entry *a = (const struct pack_midx_entry *)_a;
 510         const struct pack_midx_entry *b = (const struct pack_midx_entry *)_b;
 511         int cmp = oidcmp(&a->oid, &b->oid);
 512
 513         if (cmp)
 514                 return cmp;
 515
 516         if (a->pack_mtime > b->pack_mtime)
 517                 return -1;
 518         else if (a->pack_mtime < b->pack_mtime)
 519                 return 1;
 520
 521         return a->pack_int_id - b->pack_int_id;
 522 }
 523
 524 static int nth_midxed_pack_midx_entry(struct multi_pack_index *m,
 525                                       struct pack_midx_entry *e,
 526                                       uint32_t pos)
 527 {
 528         if (pos >= m->num_objects)
 529                 return 1;
 530
 531         nth_midxed_object_oid(&e->oid, m, pos);
 532         e->pack_int_id = nth_midxed_pack_int_id(m, pos);
 533         e->offset = nth_midxed_offset(m, pos);
 534
 535         /* consider objects in midx to be from "old" packs */
 536         e->pack_mtime = 0;
 537         return 0;
 538 }
 539
 540 static void fill_pack_entry(uint32_t pack_int_id,
 541                             struct packed_git *p,
 542                             uint32_t cur_object,
 543                             struct pack_midx_entry *entry)
 544 {
 545         if (nth_packed_object_id(&entry->oid, p, cur_object) < 0)
 546                 die(_("failed to locate object %d in packfile"), cur_object);
 547
 548         entry->pack_int_id = pack_int_id;
 549         entry->pack_mtime = p->mtime;
 550
 551         entry->offset = nth_packed_object_offset(p, cur_object);
 552 }
 553
 554 /*
 555  * It is possible to artificially get into a state where there are many
 556  * duplicate copies of objects. That can create high memory pressure if
 557  * we are to create a list of all objects before de-duplication. To reduce
 558  * this memory pressure without a significant performance drop, automatically
 559  * group objects by the first byte of their object id. Use the IDX fanout
 560  * tables to group the data, copy to a local array, then sort.
 561  *
 562  * Copy only the de-duplicated entries (selected by most-recent modified time
 563  * of a packfile containing the object).
 564  */
 565 static struct pack_midx_entry *get_sorted_entries(struct multi_pack_index *m,
 566                                                   struct pack_info *info,
 567                                                   uint32_t nr_packs,
 568                                                   uint32_t *nr_objects)
 569 {
 570         uint32_t cur_fanout, cur_pack, cur_object;
 571         uint32_t alloc_fanout, alloc_objects, total_objects = 0;
 572         struct pack_midx_entry *entries_by_fanout = NULL;
 573         struct pack_midx_entry *deduplicated_entries = NULL;
 574         uint32_t start_pack = m ? m->num_packs : 0;
 575
 576         for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++)
 577                 total_objects += info[cur_pack].p->num_objects;
 578
 579         /*
 580          * As we de-duplicate by fanout value, we expect the fanout
 581          * slices to be evenly distributed, with some noise. Hence,
 582          * allocate slightly more than one 256th.
 583          */
 584         alloc_objects = alloc_fanout = total_objects > 3200 ? total_objects / 200 : 16;
 585
 586         ALLOC_ARRAY(entries_by_fanout, alloc_fanout);
 587         ALLOC_ARRAY(deduplicated_entries, alloc_objects);
 588         *nr_objects = 0;
 589
 590         for (cur_fanout = 0; cur_fanout < 256; cur_fanout++) {
 591                 uint32_t nr_fanout = 0;
 592
 593                 if (m) {
 594                         uint32_t start = 0, end;
 595
 596                         if (cur_fanout)
 597                                 start = ntohl(m->chunk_oid_fanout[cur_fanout - 1]);
 598                         end = ntohl(m->chunk_oid_fanout[cur_fanout]);
 599
 600                         for (cur_object = start; cur_object < end; cur_object++) {
 601                                 ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
 602                                 nth_midxed_pack_midx_entry(m,
 603                                                            &entries_by_fanout[nr_fanout],
 604                                                            cur_object);
 605                                 nr_fanout++;
 606                         }
 607                 }
 608
 609                 for (cur_pack = start_pack; cur_pack < nr_packs; cur_pack++) {
 610                         uint32_t start = 0, end;
 611
 612                         if (cur_fanout)
 613                                 start = get_pack_fanout(info[cur_pack].p, cur_fanout - 1);
 614                         end = get_pack_fanout(info[cur_pack].p, cur_fanout);
 615
 616                         for (cur_object = start; cur_object < end; cur_object++) {
 617                                 ALLOC_GROW(entries_by_fanout, nr_fanout + 1, alloc_fanout);
 618                                 fill_pack_entry(cur_pack, info[cur_pack].p, cur_object, &entries_by_fanout[nr_fanout]);
 619                                 nr_fanout++;
 620                         }
 621                 }
 622
 623                 QSORT(entries_by_fanout, nr_fanout, midx_oid_compare);
 624
 625                 /*
 626                  * The batch is now sorted by OID and then mtime (descending).
 627                  * Take only the first duplicate.
 628                  */
 629                 for (cur_object = 0; cur_object < nr_fanout; cur_object++) {
 630                         if (cur_object && oideq(&entries_by_fanout[cur_object - 1].oid,
 631                                                 &entries_by_fanout[cur_object].oid))
 632                                 continue;
 633
 634                         ALLOC_GROW(deduplicated_entries, *nr_objects + 1, alloc_objects);
 635                         memcpy(&deduplicated_entries[*nr_objects],
 636                                &entries_by_fanout[cur_object],
 637                                sizeof(struct pack_midx_entry));
 638                         (*nr_objects)++;
 639                 }
 640         }
 641
 642         free(entries_by_fanout);
 643         return deduplicated_entries;
 644 }
 645
 646 static size_t write_midx_pack_names(struct hashfile *f,
 647                                     struct pack_info *info,
 648                                     uint32_t num_packs)
 649 {
 650         uint32_t i;
 651         unsigned char padding[MIDX_CHUNK_ALIGNMENT];
 652         size_t written = 0;
 653
 654         for (i = 0; i < num_packs; i++) {
 655                 size_t writelen;
 656
 657                 if (info[i].expired)
 658                         continue;
 659
 660                 if (i && strcmp(info[i].pack_name, info[i - 1].pack_name) <= 0)
 661                         BUG("incorrect pack-file order: %s before %s",
 662                             info[i - 1].pack_name,
 663                             info[i].pack_name);
 664
 665                 writelen = strlen(info[i].pack_name) + 1;
 666                 hashwrite(f, info[i].pack_name, writelen);
 667                 written += writelen;
 668         }
 669
 670         /* add padding to be aligned */
 671         i = MIDX_CHUNK_ALIGNMENT - (written % MIDX_CHUNK_ALIGNMENT);
 672         if (i < MIDX_CHUNK_ALIGNMENT) {
 673                 memset(padding, 0, sizeof(padding));
 674                 hashwrite(f, padding, i);
 675                 written += i;
 676         }
 677
 678         return written;
 679 }
 680
 681 static size_t write_midx_oid_fanout(struct hashfile *f,
 682                                     struct pack_midx_entry *objects,
 683                                     uint32_t nr_objects)
 684 {
 685         struct pack_midx_entry *list = objects;
 686         struct pack_midx_entry *last = objects + nr_objects;
 687         uint32_t count = 0;
 688         uint32_t i;
 689
 690         /*
 691         * Write the first-level table (the list is sorted,
 692         * but we use a 256-entry lookup to be able to avoid
 693         * having to do eight extra binary search iterations).
 694         */
 695         for (i = 0; i < 256; i++) {
 696                 struct pack_midx_entry *next = list;
 697
 698                 while (next < last && next->oid.hash[0] == i) {
 699                         count++;
 700                         next++;
 701                 }
 702
 703                 hashwrite_be32(f, count);
 704                 list = next;
 705         }
 706
 707         return MIDX_CHUNK_FANOUT_SIZE;
 708 }
 709
 710 static size_t write_midx_oid_lookup(struct hashfile *f, unsigned char hash_len,
 711                                     struct pack_midx_entry *objects,
 712                                     uint32_t nr_objects)
 713 {
 714         struct pack_midx_entry *list = objects;
 715         uint32_t i;
 716         size_t written = 0;
 717
 718         for (i = 0; i < nr_objects; i++) {
 719                 struct pack_midx_entry *obj = list++;
 720
 721                 if (i < nr_objects - 1) {
 722                         struct pack_midx_entry *next = list;
 723                         if (oidcmp(&obj->oid, &next->oid) >= 0)
 724                                 BUG("OIDs not in order: %s >= %s",
 725                                     oid_to_hex(&obj->oid),
 726                                     oid_to_hex(&next->oid));
 727                 }
 728
 729                 hashwrite(f, obj->oid.hash, (int)hash_len);
 730                 written += hash_len;
 731         }
 732
 733         return written;
 734 }
 735
 736 static size_t write_midx_object_offsets(struct hashfile *f, int large_offset_needed,
 737                                         uint32_t *perm,
 738                                         struct pack_midx_entry *objects, uint32_t nr_objects)
 739 {
 740         struct pack_midx_entry *list = objects;
 741         uint32_t i, nr_large_offset = 0;
 742         size_t written = 0;
 743
 744         for (i = 0; i < nr_objects; i++) {
 745                 struct pack_midx_entry *obj = list++;
 746
 747                 if (perm[obj->pack_int_id] == PACK_EXPIRED)
 748                         BUG("object %s is in an expired pack with int-id %d",
 749                             oid_to_hex(&obj->oid),
 750                             obj->pack_int_id);
 751
 752                 hashwrite_be32(f, perm[obj->pack_int_id]);
 753
 754                 if (large_offset_needed && obj->offset >> 31)
 755                         hashwrite_be32(f, MIDX_LARGE_OFFSET_NEEDED | nr_large_offset++);
 756                 else if (!large_offset_needed && obj->offset >> 32)
 757                         BUG("object %s requires a large offset (%"PRIx64") but the MIDX is not writing large offsets!",
 758                             oid_to_hex(&obj->oid),
 759                             obj->offset);
 760                 else
 761                         hashwrite_be32(f, (uint32_t)obj->offset);
 762
 763                 written += MIDX_CHUNK_OFFSET_WIDTH;
 764         }
 765
 766         return written;
 767 }
 768
 769 static size_t write_midx_large_offsets(struct hashfile *f, uint32_t nr_large_offset,
 770                                        struct pack_midx_entry *objects, uint32_t nr_objects)
 771 {
 772         struct pack_midx_entry *list = objects, *end = objects + nr_objects;
 773         size_t written = 0;
 774
 775         while (nr_large_offset) {
 776                 struct pack_midx_entry *obj;
 777                 uint64_t offset;
 778
 779                 if (list >= end)
 780                         BUG("too many large-offset objects");
 781
 782                 obj = list++;
 783                 offset = obj->offset;
 784
 785                 if (!(offset >> 31))
 786                         continue;
 787
 788                 written += hashwrite_be64(f, offset);
 789
 790                 nr_large_offset--;
 791         }
 792
 793         return written;
 794 }
 795
 796 static int write_midx_internal(const char *object_dir, struct multi_pack_index *m,
 797                                struct string_list *packs_to_drop, unsigned flags)
 798 {
 799         unsigned char cur_chunk, num_chunks = 0;
 800         char *midx_name;
 801         uint32_t i;
 802         struct hashfile *f = NULL;
 803         struct lock_file lk;
 804         struct pack_list packs;
 805         uint32_t *pack_perm = NULL;
 806         uint64_t written = 0;
 807         uint32_t chunk_ids[MIDX_MAX_CHUNKS + 1];
 808         uint64_t chunk_offsets[MIDX_MAX_CHUNKS + 1];
 809         uint32_t nr_entries, num_large_offsets = 0;
 810         struct pack_midx_entry *entries = NULL;
 811         struct progress *progress = NULL;
 812         int large_offsets_needed = 0;
 813         int pack_name_concat_len = 0;
 814         int dropped_packs = 0;
 815         int result = 0;
 816
 817         midx_name = get_midx_filename(object_dir);
 818         if (safe_create_leading_directories(midx_name))
 819                 die_errno(_("unable to create leading directories of %s"),
 820                           midx_name);
 821
 822         if (m)
 823                 packs.m = m;
 824         else
 825                 packs.m = load_multi_pack_index(object_dir, 1);
 826
 827         packs.nr = 0;
 828         packs.alloc = packs.m ? packs.m->num_packs : 16;
 829         packs.info = NULL;
 830         ALLOC_ARRAY(packs.info, packs.alloc);
 831
 832         if (packs.m) {
 833                 for (i = 0; i < packs.m->num_packs; i++) {
 834                         ALLOC_GROW(packs.info, packs.nr + 1, packs.alloc);
 835
 836                         packs.info[packs.nr].orig_pack_int_id = i;
 837                         packs.info[packs.nr].pack_name = xstrdup(packs.m->pack_names[i]);
 838                         packs.info[packs.nr].p = NULL;
 839                         packs.info[packs.nr].expired = 0;
 840                         packs.nr++;
 841                 }
 842         }
 843
 844         packs.pack_paths_checked = 0;
 845         if (flags & MIDX_PROGRESS)
 846                 packs.progress = start_delayed_progress(_("Adding packfiles to multi-pack-index"), 0);
 847         else
 848                 packs.progress = NULL;
 849
 850         for_each_file_in_pack_dir(object_dir, add_pack_to_midx, &packs);
 851         stop_progress(&packs.progress);
 852
 853         if (packs.m && packs.nr == packs.m->num_packs && !packs_to_drop)
 854                 goto cleanup;
 855
 856         entries = get_sorted_entries(packs.m, packs.info, packs.nr, &nr_entries);
 857
 858         for (i = 0; i < nr_entries; i++) {
 859                 if (entries[i].offset > 0x7fffffff)
 860                         num_large_offsets++;
 861                 if (entries[i].offset > 0xffffffff)
 862                         large_offsets_needed = 1;
 863         }
 864
 865         QSORT(packs.info, packs.nr, pack_info_compare);
 866
 867         if (packs_to_drop && packs_to_drop->nr) {
 868                 int drop_index = 0;
 869                 int missing_drops = 0;
 870
 871                 for (i = 0; i < packs.nr && drop_index < packs_to_drop->nr; i++) {
 872                         int cmp = strcmp(packs.info[i].pack_name,
 873                                          packs_to_drop->items[drop_index].string);
 874
 875                         if (!cmp) {
 876                                 drop_index++;
 877                                 packs.info[i].expired = 1;
 878                         } else if (cmp > 0) {
 879                                 error(_("did not see pack-file %s to drop"),
 880                                       packs_to_drop->items[drop_index].string);
 881                                 drop_index++;
 882                                 missing_drops++;
 883                                 i--;
 884                         } else {
 885                                 packs.info[i].expired = 0;
 886                         }
 887                 }
 888
 889                 if (missing_drops) {
 890                         result = 1;
 891                         goto cleanup;
 892                 }
 893         }
 894
 895         /*
 896          * pack_perm stores a permutation between pack-int-ids from the
 897          * previous multi-pack-index to the new one we are writing:
 898          *
 899          * pack_perm[old_id] = new_id
 900          */
 901         ALLOC_ARRAY(pack_perm, packs.nr);
 902         for (i = 0; i < packs.nr; i++) {
 903                 if (packs.info[i].expired) {
 904                         dropped_packs++;
 905                         pack_perm[packs.info[i].orig_pack_int_id] = PACK_EXPIRED;
 906                 } else {
 907                         pack_perm[packs.info[i].orig_pack_int_id] = i - dropped_packs;
 908                 }
 909         }
 910
 911         for (i = 0; i < packs.nr; i++) {
 912                 if (!packs.info[i].expired)
 913                         pack_name_concat_len += strlen(packs.info[i].pack_name) + 1;
 914         }
 915
 916         if (pack_name_concat_len % MIDX_CHUNK_ALIGNMENT)
 917                 pack_name_concat_len += MIDX_CHUNK_ALIGNMENT -
 918                                         (pack_name_concat_len % MIDX_CHUNK_ALIGNMENT);
 919
 920         hold_lock_file_for_update(&lk, midx_name, LOCK_DIE_ON_ERROR);
 921         f = hashfd(get_lock_file_fd(&lk), get_lock_file_path(&lk));
 922         FREE_AND_NULL(midx_name);
 923
 924         if (packs.m)
 925                 close_midx(packs.m);
 926
 927         cur_chunk = 0;
 928         num_chunks = large_offsets_needed ? 5 : 4;
 929
 930         if (packs.nr - dropped_packs == 0) {
 931                 error(_("no pack files to index."));
 932                 result = 1;
 933                 goto cleanup;
 934         }
 935
 936         written = write_midx_header(f, num_chunks, packs.nr - dropped_packs);
 937
 938         chunk_ids[cur_chunk] = MIDX_CHUNKID_PACKNAMES;
 939         chunk_offsets[cur_chunk] = written + (num_chunks + 1) * MIDX_CHUNKLOOKUP_WIDTH;
 940
 941         cur_chunk++;
 942         chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDFANOUT;
 943         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + pack_name_concat_len;
 944
 945         cur_chunk++;
 946         chunk_ids[cur_chunk] = MIDX_CHUNKID_OIDLOOKUP;
 947         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + MIDX_CHUNK_FANOUT_SIZE;
 948
 949         cur_chunk++;
 950         chunk_ids[cur_chunk] = MIDX_CHUNKID_OBJECTOFFSETS;
 951         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * the_hash_algo->rawsz;
 952
 953         cur_chunk++;
 954         chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] + nr_entries * MIDX_CHUNK_OFFSET_WIDTH;
 955         if (large_offsets_needed) {
 956                 chunk_ids[cur_chunk] = MIDX_CHUNKID_LARGEOFFSETS;
 957
 958                 cur_chunk++;
 959                 chunk_offsets[cur_chunk] = chunk_offsets[cur_chunk - 1] +
 960                                            num_large_offsets * MIDX_CHUNK_LARGE_OFFSET_WIDTH;
 961         }
 962
 963         chunk_ids[cur_chunk] = 0;
 964
 965         for (i = 0; i <= num_chunks; i++) {
 966                 if (i && chunk_offsets[i] < chunk_offsets[i - 1])
 967                         BUG("incorrect chunk offsets: %"PRIu64" before %"PRIu64,
 968                             chunk_offsets[i - 1],
 969                             chunk_offsets[i]);
 970
 971                 if (chunk_offsets[i] % MIDX_CHUNK_ALIGNMENT)
 972                         BUG("chunk offset %"PRIu64" is not properly aligned",
 973                             chunk_offsets[i]);
 974
 975                 hashwrite_be32(f, chunk_ids[i]);
 976                 hashwrite_be64(f, chunk_offsets[i]);
 977
 978                 written += MIDX_CHUNKLOOKUP_WIDTH;
 979         }
 980
 981         if (flags & MIDX_PROGRESS)
 982                 progress = start_delayed_progress(_("Writing chunks to multi-pack-index"),
 983                                           num_chunks);
 984         for (i = 0; i < num_chunks; i++) {
 985                 if (written != chunk_offsets[i])
 986                         BUG("incorrect chunk offset (%"PRIu64" != %"PRIu64") for chunk id %"PRIx32,
 987                             chunk_offsets[i],
 988                             written,
 989                             chunk_ids[i]);
 990
 991                 switch (chunk_ids[i]) {
 992                         case MIDX_CHUNKID_PACKNAMES:
 993                                 written += write_midx_pack_names(f, packs.info, packs.nr);
 994                                 break;
 995
 996                         case MIDX_CHUNKID_OIDFANOUT:
 997                                 written += write_midx_oid_fanout(f, entries, nr_entries);
 998                                 break;
 999
1000                         case MIDX_CHUNKID_OIDLOOKUP:
1001                                 written += write_midx_oid_lookup(f, the_hash_algo->rawsz, entries, nr_entries);
1002                                 break;
1003
1004                         case MIDX_CHUNKID_OBJECTOFFSETS:
1005                                 written += write_midx_object_offsets(f, large_offsets_needed, pack_perm, entries, nr_entries);
1006                                 break;
1007
1008                         case MIDX_CHUNKID_LARGEOFFSETS:
1009                                 written += write_midx_large_offsets(f, num_large_offsets, entries, nr_entries);
1010                                 break;
1011
1012                         default:
1013                                 BUG("trying to write unknown chunk id %"PRIx32,
1014                                     chunk_ids[i]);
1015                 }
1016
1017                 display_progress(progress, i + 1);
1018         }
1019         stop_progress(&progress);
1020
1021         if (written != chunk_offsets[num_chunks])
1022                 BUG("incorrect final offset %"PRIu64" != %"PRIu64,
1023                     written,
1024                     chunk_offsets[num_chunks]);
1025
1026         finalize_hashfile(f, NULL, CSUM_FSYNC | CSUM_HASH_IN_STREAM);
1027         commit_lock_file(&lk);
1028
1029 cleanup:
1030         for (i = 0; i < packs.nr; i++) {
1031                 if (packs.info[i].p) {
1032                         close_pack(packs.info[i].p);
1033                         free(packs.info[i].p);
1034                 }
1035                 free(packs.info[i].pack_name);
1036         }
1037
1038         free(packs.info);
1039         free(entries);
1040         free(pack_perm);
1041         free(midx_name);
1042         return result;
1043 }
1044
1045 int write_midx_file(const char *object_dir, unsigned flags)
1046 {
1047         return write_midx_internal(object_dir, NULL, NULL, flags);
1048 }
1049
1050 void clear_midx_file(struct repository *r)
1051 {
1052         char *midx = get_midx_filename(r->objects->odb->path);
1053
1054         if (r->objects && r->objects->multi_pack_index) {
1055                 close_midx(r->objects->multi_pack_index);
1056                 r->objects->multi_pack_index = NULL;
1057         }
1058
1059         if (remove_path(midx))
1060                 die(_("failed to clear multi-pack-index at %s"), midx);
1061
1062         free(midx);
1063 }
1064
1065 static int verify_midx_error;
1066
1067 static void midx_report(const char *fmt, ...)
1068 {
1069         va_list ap;
1070         verify_midx_error = 1;
1071         va_start(ap, fmt);
1072         vfprintf(stderr, fmt, ap);
1073         fprintf(stderr, "\n");
1074         va_end(ap);
1075 }
1076
1077 struct pair_pos_vs_id
1078 {
1079         uint32_t pos;
1080         uint32_t pack_int_id;
1081 };
1082
1083 static int compare_pair_pos_vs_id(const void *_a, const void *_b)
1084 {
1085         struct pair_pos_vs_id *a = (struct pair_pos_vs_id *)_a;
1086         struct pair_pos_vs_id *b = (struct pair_pos_vs_id *)_b;
1087
1088         return b->pack_int_id - a->pack_int_id;
1089 }
1090
1091 /*
1092  * Limit calls to display_progress() for performance reasons.
1093  * The interval here was arbitrarily chosen.
1094  */
1095 #define SPARSE_PROGRESS_INTERVAL (1 << 12)
1096 #define midx_display_sparse_progress(progress, n) \
1097         do { \
1098                 uint64_t _n = (n); \
1099                 if ((_n & (SPARSE_PROGRESS_INTERVAL - 1)) == 0) \
1100                         display_progress(progress, _n); \
1101         } while (0)
1102
1103 int verify_midx_file(struct repository *r, const char *object_dir, unsigned flags)
1104 {
1105         struct pair_pos_vs_id *pairs = NULL;
1106         uint32_t i;
1107         struct progress *progress = NULL;
1108         struct multi_pack_index *m = load_multi_pack_index(object_dir, 1);
1109         verify_midx_error = 0;
1110
1111         if (!m) {
1112                 int result = 0;
1113                 struct stat sb;
1114                 char *filename = get_midx_filename(object_dir);
1115                 if (!stat(filename, &sb)) {
1116                         error(_("multi-pack-index file exists, but failed to parse"));
1117                         result = 1;
1118                 }
1119                 free(filename);
1120                 return result;
1121         }
1122
1123         if (flags & MIDX_PROGRESS)
1124                 progress = start_delayed_progress(_("Looking for referenced packfiles"),
1125                                           m->num_packs);
1126         for (i = 0; i < m->num_packs; i++) {
1127                 if (prepare_midx_pack(r, m, i))
1128                         midx_report("failed to load pack in position %d", i);
1129
1130                 display_progress(progress, i + 1);
1131         }
1132         stop_progress(&progress);
1133
1134         for (i = 0; i < 255; i++) {
1135                 uint32_t oid_fanout1 = ntohl(m->chunk_oid_fanout[i]);
1136                 uint32_t oid_fanout2 = ntohl(m->chunk_oid_fanout[i + 1]);
1137
1138                 if (oid_fanout1 > oid_fanout2)
1139                         midx_report(_("oid fanout out of order: fanout[%d] = %"PRIx32" > %"PRIx32" = fanout[%d]"),
1140                                     i, oid_fanout1, oid_fanout2, i + 1);
1141         }
1142
1143         if (m->num_objects == 0) {
1144                 midx_report(_("the midx contains no oid"));
1145                 /*
1146                  * Remaining tests assume that we have objects, so we can
1147                  * return here.
1148                  */
1149                 return verify_midx_error;
1150         }
1151
1152         if (flags & MIDX_PROGRESS)
1153                 progress = start_sparse_progress(_("Verifying OID order in multi-pack-index"),
1154                                                  m->num_objects - 1);
1155         for (i = 0; i < m->num_objects - 1; i++) {
1156                 struct object_id oid1, oid2;
1157
1158                 nth_midxed_object_oid(&oid1, m, i);
1159                 nth_midxed_object_oid(&oid2, m, i + 1);
1160
1161                 if (oidcmp(&oid1, &oid2) >= 0)
1162                         midx_report(_("oid lookup out of order: oid[%d] = %s >= %s = oid[%d]"),
1163                                     i, oid_to_hex(&oid1), oid_to_hex(&oid2), i + 1);
1164
1165                 midx_display_sparse_progress(progress, i + 1);
1166         }
1167         stop_progress(&progress);
1168
1169         /*
1170          * Create an array mapping each object to its packfile id.  Sort it
1171          * to group the objects by packfile.  Use this permutation to visit
1172          * each of the objects and only require 1 packfile to be open at a
1173          * time.
1174          */
1175         ALLOC_ARRAY(pairs, m->num_objects);
1176         for (i = 0; i < m->num_objects; i++) {
1177                 pairs[i].pos = i;
1178                 pairs[i].pack_int_id = nth_midxed_pack_int_id(m, i);
1179         }
1180
1181         if (flags & MIDX_PROGRESS)
1182                 progress = start_sparse_progress(_("Sorting objects by packfile"),
1183                                                  m->num_objects);
1184         display_progress(progress, 0); /* TODO: Measure QSORT() progress */
1185         QSORT(pairs, m->num_objects, compare_pair_pos_vs_id);
1186         stop_progress(&progress);
1187
1188         if (flags & MIDX_PROGRESS)
1189                 progress = start_sparse_progress(_("Verifying object offsets"), m->num_objects);
1190         for (i = 0; i < m->num_objects; i++) {
1191                 struct object_id oid;
1192                 struct pack_entry e;
1193                 off_t m_offset, p_offset;
1194
1195                 if (i > 0 && pairs[i-1].pack_int_id != pairs[i].pack_int_id &&
1196                     m->packs[pairs[i-1].pack_int_id])
1197                 {
1198                         close_pack_fd(m->packs[pairs[i-1].pack_int_id]);
1199                         close_pack_index(m->packs[pairs[i-1].pack_int_id]);
1200                 }
1201
1202                 nth_midxed_object_oid(&oid, m, pairs[i].pos);
1203
1204                 if (!fill_midx_entry(r, &oid, &e, m)) {
1205                         midx_report(_("failed to load pack entry for oid[%d] = %s"),
1206                                     pairs[i].pos, oid_to_hex(&oid));
1207                         continue;
1208                 }
1209
1210                 if (open_pack_index(e.p)) {
1211                         midx_report(_("failed to load pack-index for packfile %s"),
1212                                     e.p->pack_name);
1213                         break;
1214                 }
1215
1216                 m_offset = e.offset;
1217                 p_offset = find_pack_entry_one(oid.hash, e.p);
1218
1219                 if (m_offset != p_offset)
1220                         midx_report(_("incorrect object offset for oid[%d] = %s: %"PRIx64" != %"PRIx64),
1221                                     pairs[i].pos, oid_to_hex(&oid), m_offset, p_offset);
1222
1223                 midx_display_sparse_progress(progress, i + 1);
1224         }
1225         stop_progress(&progress);
1226
1227         free(pairs);
1228
1229         return verify_midx_error;
1230 }
1231
1232 int expire_midx_packs(struct repository *r, const char *object_dir, unsigned flags)
1233 {
1234         uint32_t i, *count, result = 0;
1235         struct string_list packs_to_drop = STRING_LIST_INIT_DUP;
1236         struct multi_pack_index *m = load_multi_pack_index(object_dir, 1);
1237         struct progress *progress = NULL;
1238
1239         if (!m)
1240                 return 0;
1241
1242         count = xcalloc(m->num_packs, sizeof(uint32_t));
1243
1244         if (flags & MIDX_PROGRESS)
1245                 progress = start_delayed_progress(_("Counting referenced objects"),
1246                                           m->num_objects);
1247         for (i = 0; i < m->num_objects; i++) {
1248                 int pack_int_id = nth_midxed_pack_int_id(m, i);
1249                 count[pack_int_id]++;
1250                 display_progress(progress, i + 1);
1251         }
1252         stop_progress(&progress);
1253
1254         if (flags & MIDX_PROGRESS)
1255                 progress = start_delayed_progress(_("Finding and deleting unreferenced packfiles"),
1256                                           m->num_packs);
1257         for (i = 0; i < m->num_packs; i++) {
1258                 char *pack_name;
1259                 display_progress(progress, i + 1);
1260
1261                 if (count[i])
1262                         continue;
1263
1264                 if (prepare_midx_pack(r, m, i))
1265                         continue;
1266
1267                 if (m->packs[i]->pack_keep)
1268                         continue;
1269
1270                 pack_name = xstrdup(m->packs[i]->pack_name);
1271                 close_pack(m->packs[i]);
1272
1273                 string_list_insert(&packs_to_drop, m->pack_names[i]);
1274                 unlink_pack_path(pack_name, 0);
1275                 free(pack_name);
1276         }
1277         stop_progress(&progress);
1278
1279         free(count);
1280
1281         if (packs_to_drop.nr)
1282                 result = write_midx_internal(object_dir, m, &packs_to_drop, flags);
1283
1284         string_list_clear(&packs_to_drop, 0);
1285         return result;
1286 }
1287
1288 struct repack_info {
1289         timestamp_t mtime;
1290         uint32_t referenced_objects;
1291         uint32_t pack_int_id;
1292 };
1293
1294 static int compare_by_mtime(const void *a_, const void *b_)
1295 {
1296         const struct repack_info *a, *b;
1297
1298         a = (const struct repack_info *)a_;
1299         b = (const struct repack_info *)b_;
1300
1301         if (a->mtime < b->mtime)
1302                 return -1;
1303         if (a->mtime > b->mtime)
1304                 return 1;
1305         return 0;
1306 }
1307
1308 static int fill_included_packs_all(struct repository *r,
1309                                    struct multi_pack_index *m,
1310                                    unsigned char *include_pack)
1311 {
1312         uint32_t i, count = 0;
1313         int pack_kept_objects = 0;
1314
1315         repo_config_get_bool(r, "repack.packkeptobjects", &pack_kept_objects);
1316
1317         for (i = 0; i < m->num_packs; i++) {
1318                 if (prepare_midx_pack(r, m, i))
1319                         continue;
1320                 if (!pack_kept_objects && m->packs[i]->pack_keep)
1321                         continue;
1322
1323                 include_pack[i] = 1;
1324                 count++;
1325         }
1326
1327         return count < 2;
1328 }
1329
1330 static int fill_included_packs_batch(struct repository *r,
1331                                      struct multi_pack_index *m,
1332                                      unsigned char *include_pack,
1333                                      size_t batch_size)
1334 {
1335         uint32_t i, packs_to_repack;
1336         size_t total_size;
1337         struct repack_info *pack_info = xcalloc(m->num_packs, sizeof(struct repack_info));
1338         int pack_kept_objects = 0;
1339
1340         repo_config_get_bool(r, "repack.packkeptobjects", &pack_kept_objects);
1341
1342         for (i = 0; i < m->num_packs; i++) {
1343                 pack_info[i].pack_int_id = i;
1344
1345                 if (prepare_midx_pack(r, m, i))
1346                         continue;
1347
1348                 pack_info[i].mtime = m->packs[i]->mtime;
1349         }
1350
1351         for (i = 0; batch_size && i < m->num_objects; i++) {
1352                 uint32_t pack_int_id = nth_midxed_pack_int_id(m, i);
1353                 pack_info[pack_int_id].referenced_objects++;
1354         }
1355
1356         QSORT(pack_info, m->num_packs, compare_by_mtime);
1357
1358         total_size = 0;
1359         packs_to_repack = 0;
1360         for (i = 0; total_size < batch_size && i < m->num_packs; i++) {
1361                 int pack_int_id = pack_info[i].pack_int_id;
1362                 struct packed_git *p = m->packs[pack_int_id];
1363                 size_t expected_size;
1364
1365                 if (!p)
1366                         continue;
1367                 if (!pack_kept_objects && p->pack_keep)
1368                         continue;
1369                 if (open_pack_index(p) || !p->num_objects)
1370                         continue;
1371
1372                 expected_size = (size_t)(p->pack_size
1373                                          * pack_info[i].referenced_objects);
1374                 expected_size /= p->num_objects;
1375
1376                 if (expected_size >= batch_size)
1377                         continue;
1378
1379                 packs_to_repack++;
1380                 total_size += expected_size;
1381                 include_pack[pack_int_id] = 1;
1382         }
1383
1384         free(pack_info);
1385
1386         if (packs_to_repack < 2)
1387                 return 1;
1388
1389         return 0;
1390 }
1391
1392 int midx_repack(struct repository *r, const char *object_dir, size_t batch_size, unsigned flags)
1393 {
1394         int result = 0;
1395         uint32_t i;
1396         unsigned char *include_pack;
1397         struct child_process cmd = CHILD_PROCESS_INIT;
1398         FILE *cmd_in;
1399         struct strbuf base_name = STRBUF_INIT;
1400         struct multi_pack_index *m = load_multi_pack_index(object_dir, 1);
1401
1402         /*
1403          * When updating the default for these configuration
1404          * variables in builtin/repack.c, these must be adjusted
1405          * to match.
1406          */
1407         int delta_base_offset = 1;
1408         int use_delta_islands = 0;
1409
1410         if (!m)
1411                 return 0;
1412
1413         include_pack = xcalloc(m->num_packs, sizeof(unsigned char));
1414
1415         if (batch_size) {
1416                 if (fill_included_packs_batch(r, m, include_pack, batch_size))
1417                         goto cleanup;
1418         } else if (fill_included_packs_all(r, m, include_pack))
1419                 goto cleanup;
1420
1421         repo_config_get_bool(r, "repack.usedeltabaseoffset", &delta_base_offset);
1422         repo_config_get_bool(r, "repack.usedeltaislands", &use_delta_islands);
1423
1424         strvec_push(&cmd.args, "pack-objects");
1425
1426         strbuf_addstr(&base_name, object_dir);
1427         strbuf_addstr(&base_name, "/pack/pack");
1428         strvec_push(&cmd.args, base_name.buf);
1429
1430         if (delta_base_offset)
1431                 strvec_push(&cmd.args, "--delta-base-offset");
1432         if (use_delta_islands)
1433                 strvec_push(&cmd.args, "--delta-islands");
1434
1435         if (flags & MIDX_PROGRESS)
1436                 strvec_push(&cmd.args, "--progress");
1437         else
1438                 strvec_push(&cmd.args, "-q");
1439
1440         strbuf_release(&base_name);
1441
1442         cmd.git_cmd = 1;
1443         cmd.in = cmd.out = -1;
1444
1445         if (start_command(&cmd)) {
1446                 error(_("could not start pack-objects"));
1447                 result = 1;
1448                 goto cleanup;
1449         }
1450
1451         cmd_in = xfdopen(cmd.in, "w");
1452
1453         for (i = 0; i < m->num_objects; i++) {
1454                 struct object_id oid;
1455                 uint32_t pack_int_id = nth_midxed_pack_int_id(m, i);
1456
1457                 if (!include_pack[pack_int_id])
1458                         continue;
1459
1460                 nth_midxed_object_oid(&oid, m, i);
1461                 fprintf(cmd_in, "%s\n", oid_to_hex(&oid));
1462         }
1463         fclose(cmd_in);
1464
1465         if (finish_command(&cmd)) {
1466                 error(_("could not finish pack-objects"));
1467                 result = 1;
1468                 goto cleanup;
1469         }
1470
1471         result = write_midx_internal(object_dir, m, NULL, flags);
1472         m = NULL;
1473
1474 cleanup:
1475         if (m)
1476                 close_midx(m);
1477         free(include_pack);
1478         return result;
1479 }