builtin/index-pack.c

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "thread-utils.h"
  13
  14 static const char index_pack_usage[] =
  15 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  16
  17 struct object_entry {
  18         struct pack_idx_entry idx;
  19         unsigned long size;
  20         unsigned int hdr_size;
  21         enum object_type type;
  22         enum object_type real_type;
  23         unsigned delta_depth;
  24         int base_object_no;
  25 };
  26
  27 union delta_base {
  28         unsigned char sha1[20];
  29         off_t offset;
  30 };
  31
  32 struct base_data {
  33         struct base_data *base;
  34         struct base_data *child;
  35         struct object_entry *obj;
  36         void *data;
  37         unsigned long size;
  38         int ref_first, ref_last;
  39         int ofs_first, ofs_last;
  40 };
  41
  42 #if !defined(NO_PTHREADS) && defined(NO_PREAD)
  43 /* NO_PREAD uses compat/pread.c, which is not thread-safe. Disable threading. */
  44 #define NO_PTHREADS
  45 #endif
  46
  47 struct thread_local {
  48 #ifndef NO_PTHREADS
  49         pthread_t thread;
  50 #endif
  51         struct base_data *base_cache;
  52         size_t base_cache_used;
  53 };
  54
  55 /*
  56  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  57  * to memcmp() only the first 20 bytes.
  58  */
  59 #define UNION_BASE_SZ   20
  60
  61 #define FLAG_LINK (1u<<20)
  62 #define FLAG_CHECKED (1u<<21)
  63
  64 struct delta_entry {
  65         union delta_base base;
  66         int obj_no;
  67 };
  68
  69 static struct object_entry *objects;
  70 static struct delta_entry *deltas;
  71 static struct thread_local nothread_data;
  72 static int nr_objects;
  73 static int nr_deltas;
  74 static int nr_resolved_deltas;
  75 static int nr_threads;
  76
  77 static int from_stdin;
  78 static int strict;
  79 static int verbose;
  80
  81 static struct progress *progress;
  82
  83 /* We always read in 4kB chunks. */
  84 static unsigned char input_buffer[4096];
  85 static unsigned int input_offset, input_len;
  86 static off_t consumed_bytes;
  87 static unsigned deepest_delta;
  88 static git_SHA_CTX input_ctx;
  89 static uint32_t input_crc32;
  90 static int input_fd, output_fd, pack_fd;
  91
  92 #ifndef NO_PTHREADS
  93
  94 static struct thread_local *thread_data;
  95 static int nr_dispatched;
  96 static int threads_active;
  97
  98 static pthread_mutex_t read_mutex;
  99 #define read_lock()             lock_mutex(&read_mutex)
 100 #define read_unlock()           unlock_mutex(&read_mutex)
 101
 102 static pthread_mutex_t counter_mutex;
 103 #define counter_lock()          lock_mutex(&counter_mutex)
 104 #define counter_unlock()        unlock_mutex(&counter_mutex)
 105
 106 static pthread_mutex_t work_mutex;
 107 #define work_lock()             lock_mutex(&work_mutex)
 108 #define work_unlock()           unlock_mutex(&work_mutex)
 109
 110 static pthread_key_t key;
 111
 112 static inline void lock_mutex(pthread_mutex_t *mutex)
 113 {
 114         if (threads_active)
 115                 pthread_mutex_lock(mutex);
 116 }
 117
 118 static inline void unlock_mutex(pthread_mutex_t *mutex)
 119 {
 120         if (threads_active)
 121                 pthread_mutex_unlock(mutex);
 122 }
 123
 124 /*
 125  * Mutex and conditional variable can't be statically-initialized on Windows.
 126  */
 127 static void init_thread(void)
 128 {
 129         init_recursive_mutex(&read_mutex);
 130         pthread_mutex_init(&counter_mutex, NULL);
 131         pthread_mutex_init(&work_mutex, NULL);
 132         pthread_key_create(&key, NULL);
 133         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 134         threads_active = 1;
 135 }
 136
 137 static void cleanup_thread(void)
 138 {
 139         if (!threads_active)
 140                 return;
 141         threads_active = 0;
 142         pthread_mutex_destroy(&read_mutex);
 143         pthread_mutex_destroy(&counter_mutex);
 144         pthread_mutex_destroy(&work_mutex);
 145         pthread_key_delete(key);
 146         free(thread_data);
 147 }
 148
 149 #else
 150
 151 #define read_lock()
 152 #define read_unlock()
 153
 154 #define counter_lock()
 155 #define counter_unlock()
 156
 157 #define work_lock()
 158 #define work_unlock()
 159
 160 #endif
 161
 162
 163 static int mark_link(struct object *obj, int type, void *data)
 164 {
 165         if (!obj)
 166                 return -1;
 167
 168         if (type != OBJ_ANY && obj->type != type)
 169                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 170
 171         obj->flags |= FLAG_LINK;
 172         return 0;
 173 }
 174
 175 /* The content of each linked object must have been checked
 176    or it must be already present in the object database */
 177 static void check_object(struct object *obj)
 178 {
 179         if (!obj)
 180                 return;
 181
 182         if (!(obj->flags & FLAG_LINK))
 183                 return;
 184
 185         if (!(obj->flags & FLAG_CHECKED)) {
 186                 unsigned long size;
 187                 int type = sha1_object_info(obj->sha1, &size);
 188                 if (type != obj->type || type <= 0)
 189                         die(_("object of unexpected type"));
 190                 obj->flags |= FLAG_CHECKED;
 191                 return;
 192         }
 193 }
 194
 195 static void check_objects(void)
 196 {
 197         unsigned i, max;
 198
 199         max = get_max_object_index();
 200         for (i = 0; i < max; i++)
 201                 check_object(get_indexed_object(i));
 202 }
 203
 204
 205 /* Discard current buffer used content. */
 206 static void flush(void)
 207 {
 208         if (input_offset) {
 209                 if (output_fd >= 0)
 210                         write_or_die(output_fd, input_buffer, input_offset);
 211                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 212                 memmove(input_buffer, input_buffer + input_offset, input_len);
 213                 input_offset = 0;
 214         }
 215 }
 216
 217 /*
 218  * Make sure at least "min" bytes are available in the buffer, and
 219  * return the pointer to the buffer.
 220  */
 221 static void *fill(int min)
 222 {
 223         if (min <= input_len)
 224                 return input_buffer + input_offset;
 225         if (min > sizeof(input_buffer))
 226                 die(Q_("cannot fill %d byte",
 227                        "cannot fill %d bytes",
 228                        min),
 229                     min);
 230         flush();
 231         do {
 232                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 233                                 sizeof(input_buffer) - input_len);
 234                 if (ret <= 0) {
 235                         if (!ret)
 236                                 die(_("early EOF"));
 237                         die_errno(_("read error on input"));
 238                 }
 239                 input_len += ret;
 240                 if (from_stdin)
 241                         display_throughput(progress, consumed_bytes + input_len);
 242         } while (input_len < min);
 243         return input_buffer;
 244 }
 245
 246 static void use(int bytes)
 247 {
 248         if (bytes > input_len)
 249                 die(_("used more bytes than were available"));
 250         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 251         input_len -= bytes;
 252         input_offset += bytes;
 253
 254         /* make sure off_t is sufficiently large not to wrap */
 255         if (signed_add_overflows(consumed_bytes, bytes))
 256                 die(_("pack too large for current definition of off_t"));
 257         consumed_bytes += bytes;
 258 }
 259
 260 static const char *open_pack_file(const char *pack_name)
 261 {
 262         if (from_stdin) {
 263                 input_fd = 0;
 264                 if (!pack_name) {
 265                         static char tmp_file[PATH_MAX];
 266                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 267                                                 "pack/tmp_pack_XXXXXX");
 268                         pack_name = xstrdup(tmp_file);
 269                 } else
 270                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 271                 if (output_fd < 0)
 272                         die_errno(_("unable to create '%s'"), pack_name);
 273                 pack_fd = output_fd;
 274         } else {
 275                 input_fd = open(pack_name, O_RDONLY);
 276                 if (input_fd < 0)
 277                         die_errno(_("cannot open packfile '%s'"), pack_name);
 278                 output_fd = -1;
 279                 pack_fd = input_fd;
 280         }
 281         git_SHA1_Init(&input_ctx);
 282         return pack_name;
 283 }
 284
 285 static void parse_pack_header(void)
 286 {
 287         struct pack_header *hdr = fill(sizeof(struct pack_header));
 288
 289         /* Header consistency check */
 290         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 291                 die(_("pack signature mismatch"));
 292         if (!pack_version_ok(hdr->hdr_version))
 293                 die("pack version %"PRIu32" unsupported",
 294                         ntohl(hdr->hdr_version));
 295
 296         nr_objects = ntohl(hdr->hdr_entries);
 297         use(sizeof(struct pack_header));
 298 }
 299
 300 static NORETURN void bad_object(unsigned long offset, const char *format,
 301                        ...) __attribute__((format (printf, 2, 3)));
 302
 303 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 304 {
 305         va_list params;
 306         char buf[1024];
 307
 308         va_start(params, format);
 309         vsnprintf(buf, sizeof(buf), format, params);
 310         va_end(params);
 311         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 312 }
 313
 314 static inline struct thread_local *get_thread_data(void)
 315 {
 316 #ifndef NO_PTHREADS
 317         if (threads_active)
 318                 return pthread_getspecific(key);
 319         assert(!threads_active &&
 320                "This should only be reached when all threads are gone");
 321 #endif
 322         return &nothread_data;
 323 }
 324
 325 #ifndef NO_PTHREADS
 326 static void set_thread_data(struct thread_local *data)
 327 {
 328         if (threads_active)
 329                 pthread_setspecific(key, data);
 330 }
 331 #endif
 332
 333 static struct base_data *alloc_base_data(void)
 334 {
 335         struct base_data *base = xmalloc(sizeof(struct base_data));
 336         memset(base, 0, sizeof(*base));
 337         base->ref_last = -1;
 338         base->ofs_last = -1;
 339         return base;
 340 }
 341
 342 static void free_base_data(struct base_data *c)
 343 {
 344         if (c->data) {
 345                 free(c->data);
 346                 c->data = NULL;
 347                 get_thread_data()->base_cache_used -= c->size;
 348         }
 349 }
 350
 351 static void prune_base_data(struct base_data *retain)
 352 {
 353         struct base_data *b;
 354         struct thread_local *data = get_thread_data();
 355         for (b = data->base_cache;
 356              data->base_cache_used > delta_base_cache_limit && b;
 357              b = b->child) {
 358                 if (b->data && b != retain)
 359                         free_base_data(b);
 360         }
 361 }
 362
 363 static void link_base_data(struct base_data *base, struct base_data *c)
 364 {
 365         if (base)
 366                 base->child = c;
 367         else
 368                 get_thread_data()->base_cache = c;
 369
 370         c->base = base;
 371         c->child = NULL;
 372         if (c->data)
 373                 get_thread_data()->base_cache_used += c->size;
 374         prune_base_data(c);
 375 }
 376
 377 static void unlink_base_data(struct base_data *c)
 378 {
 379         struct base_data *base = c->base;
 380         if (base)
 381                 base->child = NULL;
 382         else
 383                 get_thread_data()->base_cache = NULL;
 384         free_base_data(c);
 385 }
 386
 387 static int is_delta_type(enum object_type type)
 388 {
 389         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 390 }
 391
 392 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 393                                enum object_type type, unsigned char *sha1)
 394 {
 395         static char fixed_buf[8192];
 396         int status;
 397         git_zstream stream;
 398         void *buf;
 399         git_SHA_CTX c;
 400         char hdr[32];
 401         int hdrlen;
 402
 403         if (!is_delta_type(type)) {
 404                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 405                 git_SHA1_Init(&c);
 406                 git_SHA1_Update(&c, hdr, hdrlen);
 407         } else
 408                 sha1 = NULL;
 409         if (type == OBJ_BLOB && size > big_file_threshold)
 410                 buf = fixed_buf;
 411         else
 412                 buf = xmalloc(size);
 413
 414         memset(&stream, 0, sizeof(stream));
 415         git_inflate_init(&stream);
 416         stream.next_out = buf;
 417         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 418
 419         do {
 420                 unsigned char *last_out = stream.next_out;
 421                 stream.next_in = fill(1);
 422                 stream.avail_in = input_len;
 423                 status = git_inflate(&stream, 0);
 424                 use(input_len - stream.avail_in);
 425                 if (sha1)
 426                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 427                 if (buf == fixed_buf) {
 428                         stream.next_out = buf;
 429                         stream.avail_out = sizeof(fixed_buf);
 430                 }
 431         } while (status == Z_OK);
 432         if (stream.total_out != size || status != Z_STREAM_END)
 433                 bad_object(offset, _("inflate returned %d"), status);
 434         git_inflate_end(&stream);
 435         if (sha1)
 436                 git_SHA1_Final(sha1, &c);
 437         return buf == fixed_buf ? NULL : buf;
 438 }
 439
 440 static void *unpack_raw_entry(struct object_entry *obj,
 441                               union delta_base *delta_base,
 442                               unsigned char *sha1)
 443 {
 444         unsigned char *p;
 445         unsigned long size, c;
 446         off_t base_offset;
 447         unsigned shift;
 448         void *data;
 449
 450         obj->idx.offset = consumed_bytes;
 451         input_crc32 = crc32(0, NULL, 0);
 452
 453         p = fill(1);
 454         c = *p;
 455         use(1);
 456         obj->type = (c >> 4) & 7;
 457         size = (c & 15);
 458         shift = 4;
 459         while (c & 0x80) {
 460                 p = fill(1);
 461                 c = *p;
 462                 use(1);
 463                 size += (c & 0x7f) << shift;
 464                 shift += 7;
 465         }
 466         obj->size = size;
 467
 468         switch (obj->type) {
 469         case OBJ_REF_DELTA:
 470                 hashcpy(delta_base->sha1, fill(20));
 471                 use(20);
 472                 break;
 473         case OBJ_OFS_DELTA:
 474                 memset(delta_base, 0, sizeof(*delta_base));
 475                 p = fill(1);
 476                 c = *p;
 477                 use(1);
 478                 base_offset = c & 127;
 479                 while (c & 128) {
 480                         base_offset += 1;
 481                         if (!base_offset || MSB(base_offset, 7))
 482                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 483                         p = fill(1);
 484                         c = *p;
 485                         use(1);
 486                         base_offset = (base_offset << 7) + (c & 127);
 487                 }
 488                 delta_base->offset = obj->idx.offset - base_offset;
 489                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 490                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 491                 break;
 492         case OBJ_COMMIT:
 493         case OBJ_TREE:
 494         case OBJ_BLOB:
 495         case OBJ_TAG:
 496                 break;
 497         default:
 498                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 499         }
 500         obj->hdr_size = consumed_bytes - obj->idx.offset;
 501
 502         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 503         obj->idx.crc32 = input_crc32;
 504         return data;
 505 }
 506
 507 static void *unpack_data(struct object_entry *obj,
 508                          int (*consume)(const unsigned char *, unsigned long, void *),
 509                          void *cb_data)
 510 {
 511         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 512         unsigned long len = obj[1].idx.offset - from;
 513         unsigned char *data, *inbuf;
 514         git_zstream stream;
 515         int status;
 516
 517         data = xmalloc(consume ? 64*1024 : obj->size);
 518         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 519
 520         memset(&stream, 0, sizeof(stream));
 521         git_inflate_init(&stream);
 522         stream.next_out = data;
 523         stream.avail_out = consume ? 64*1024 : obj->size;
 524
 525         do {
 526                 unsigned char *last_out = stream.next_out;
 527                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 528                 n = pread(pack_fd, inbuf, n, from);
 529                 if (n < 0)
 530                         die_errno(_("cannot pread pack file"));
 531                 if (!n)
 532                         die(Q_("premature end of pack file, %lu byte missing",
 533                                "premature end of pack file, %lu bytes missing",
 534                                len),
 535                             len);
 536                 from += n;
 537                 len -= n;
 538                 stream.next_in = inbuf;
 539                 stream.avail_in = n;
 540                 status = git_inflate(&stream, 0);
 541                 if (consume) {
 542                         if (consume(last_out, stream.next_out - last_out, cb_data)) {
 543                                 free(inbuf);
 544                                 free(data);
 545                                 return NULL;
 546                         }
 547                         stream.next_out = data;
 548                         stream.avail_out = 64*1024;
 549                 }
 550         } while (len && status == Z_OK && !stream.avail_in);
 551
 552         /* This has been inflated OK when first encountered, so... */
 553         if (status != Z_STREAM_END || stream.total_out != obj->size)
 554                 die(_("serious inflate inconsistency"));
 555
 556         git_inflate_end(&stream);
 557         free(inbuf);
 558         if (consume) {
 559                 free(data);
 560                 data = NULL;
 561         }
 562         return data;
 563 }
 564
 565 static void *get_data_from_pack(struct object_entry *obj)
 566 {
 567         return unpack_data(obj, NULL, NULL);
 568 }
 569
 570 static int compare_delta_bases(const union delta_base *base1,
 571                                const union delta_base *base2,
 572                                enum object_type type1,
 573                                enum object_type type2)
 574 {
 575         int cmp = type1 - type2;
 576         if (cmp)
 577                 return cmp;
 578         return memcmp(base1, base2, UNION_BASE_SZ);
 579 }
 580
 581 static int find_delta(const union delta_base *base, enum object_type type)
 582 {
 583         int first = 0, last = nr_deltas;
 584
 585         while (first < last) {
 586                 int next = (first + last) / 2;
 587                 struct delta_entry *delta = &deltas[next];
 588                 int cmp;
 589
 590                 cmp = compare_delta_bases(base, &delta->base,
 591                                           type, objects[delta->obj_no].type);
 592                 if (!cmp)
 593                         return next;
 594                 if (cmp < 0) {
 595                         last = next;
 596                         continue;
 597                 }
 598                 first = next+1;
 599         }
 600         return -first-1;
 601 }
 602
 603 static void find_delta_children(const union delta_base *base,
 604                                 int *first_index, int *last_index,
 605                                 enum object_type type)
 606 {
 607         int first = find_delta(base, type);
 608         int last = first;
 609         int end = nr_deltas - 1;
 610
 611         if (first < 0) {
 612                 *first_index = 0;
 613                 *last_index = -1;
 614                 return;
 615         }
 616         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 617                 --first;
 618         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 619                 ++last;
 620         *first_index = first;
 621         *last_index = last;
 622 }
 623
 624 static void sha1_object(const void *data, struct object_entry *obj_entry,
 625                         unsigned long size, enum object_type type,
 626                         const unsigned char *sha1)
 627 {
 628         void *new_data = NULL;
 629
 630         assert(data || obj_entry);
 631
 632         read_lock();
 633         if (has_sha1_file(sha1)) {
 634                 void *has_data;
 635                 enum object_type has_type;
 636                 unsigned long has_size;
 637                 if (!data)
 638                         data = new_data = get_data_from_pack(obj_entry);
 639                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 640                 read_unlock();
 641                 if (!has_data)
 642                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 643                 if (size != has_size || type != has_type ||
 644                     memcmp(data, has_data, size) != 0)
 645                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 646                 free(has_data);
 647         } else
 648                 read_unlock();
 649
 650         if (strict) {
 651                 read_lock();
 652                 if (type == OBJ_BLOB) {
 653                         struct blob *blob = lookup_blob(sha1);
 654                         if (blob)
 655                                 blob->object.flags |= FLAG_CHECKED;
 656                         else
 657                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 658                 } else {
 659                         struct object *obj;
 660                         int eaten;
 661                         void *buf = (void *) data;
 662
 663                         if (!buf)
 664                                 buf = new_data = get_data_from_pack(obj_entry);
 665
 666                         /*
 667                          * we do not need to free the memory here, as the
 668                          * buf is deleted by the caller.
 669                          */
 670                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 671                         if (!obj)
 672                                 die(_("invalid %s"), typename(type));
 673                         if (fsck_object(obj, 1, fsck_error_function))
 674                                 die(_("Error in object"));
 675                         if (fsck_walk(obj, mark_link, NULL))
 676                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 677
 678                         if (obj->type == OBJ_TREE) {
 679                                 struct tree *item = (struct tree *) obj;
 680                                 item->buffer = NULL;
 681                         }
 682                         if (obj->type == OBJ_COMMIT) {
 683                                 struct commit *commit = (struct commit *) obj;
 684                                 commit->buffer = NULL;
 685                         }
 686                         obj->flags |= FLAG_CHECKED;
 687                 }
 688                 read_unlock();
 689         }
 690
 691         free(new_data);
 692 }
 693
 694 /*
 695  * This function is part of find_unresolved_deltas(). There are two
 696  * walkers going in the opposite ways.
 697  *
 698  * The first one in find_unresolved_deltas() traverses down from
 699  * parent node to children, deflating nodes along the way. However,
 700  * memory for deflated nodes is limited by delta_base_cache_limit, so
 701  * at some point parent node's deflated content may be freed.
 702  *
 703  * The second walker is this function, which goes from current node up
 704  * to top parent if necessary to deflate the node. In normal
 705  * situation, its parent node would be already deflated, so it just
 706  * needs to apply delta.
 707  *
 708  * In the worst case scenario, parent node is no longer deflated because
 709  * we're running out of delta_base_cache_limit; we need to re-deflate
 710  * parents, possibly up to the top base.
 711  *
 712  * All deflated objects here are subject to be freed if we exceed
 713  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 714  * just need to make sure the last node is not freed.
 715  */
 716 static void *get_base_data(struct base_data *c)
 717 {
 718         if (!c->data) {
 719                 struct object_entry *obj = c->obj;
 720                 struct base_data **delta = NULL;
 721                 int delta_nr = 0, delta_alloc = 0;
 722
 723                 while (is_delta_type(c->obj->type) && !c->data) {
 724                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 725                         delta[delta_nr++] = c;
 726                         c = c->base;
 727                 }
 728                 if (!delta_nr) {
 729                         c->data = get_data_from_pack(obj);
 730                         c->size = obj->size;
 731                         get_thread_data()->base_cache_used += c->size;
 732                         prune_base_data(c);
 733                 }
 734                 for (; delta_nr > 0; delta_nr--) {
 735                         void *base, *raw;
 736                         c = delta[delta_nr - 1];
 737                         obj = c->obj;
 738                         base = get_base_data(c->base);
 739                         raw = get_data_from_pack(obj);
 740                         c->data = patch_delta(
 741                                 base, c->base->size,
 742                                 raw, obj->size,
 743                                 &c->size);
 744                         free(raw);
 745                         if (!c->data)
 746                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 747                         get_thread_data()->base_cache_used += c->size;
 748                         prune_base_data(c);
 749                 }
 750                 free(delta);
 751         }
 752         return c->data;
 753 }
 754
 755 static void resolve_delta(struct object_entry *delta_obj,
 756                           struct base_data *base, struct base_data *result)
 757 {
 758         void *base_data, *delta_data;
 759
 760         delta_obj->real_type = base->obj->real_type;
 761         delta_obj->delta_depth = base->obj->delta_depth + 1;
 762         if (deepest_delta < delta_obj->delta_depth)
 763                 deepest_delta = delta_obj->delta_depth;
 764         delta_obj->base_object_no = base->obj - objects;
 765         delta_data = get_data_from_pack(delta_obj);
 766         base_data = get_base_data(base);
 767         result->obj = delta_obj;
 768         result->data = patch_delta(base_data, base->size,
 769                                    delta_data, delta_obj->size, &result->size);
 770         free(delta_data);
 771         if (!result->data)
 772                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 773         hash_sha1_file(result->data, result->size,
 774                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 775         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 776                     delta_obj->idx.sha1);
 777         counter_lock();
 778         nr_resolved_deltas++;
 779         counter_unlock();
 780 }
 781
 782 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 783                                                   struct base_data *prev_base)
 784 {
 785         if (base->ref_last == -1 && base->ofs_last == -1) {
 786                 union delta_base base_spec;
 787
 788                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 789                 find_delta_children(&base_spec,
 790                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 791
 792                 memset(&base_spec, 0, sizeof(base_spec));
 793                 base_spec.offset = base->obj->idx.offset;
 794                 find_delta_children(&base_spec,
 795                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 796
 797                 if (base->ref_last == -1 && base->ofs_last == -1) {
 798                         free(base->data);
 799                         return NULL;
 800                 }
 801
 802                 link_base_data(prev_base, base);
 803         }
 804
 805         if (base->ref_first <= base->ref_last) {
 806                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 807                 struct base_data *result = alloc_base_data();
 808
 809                 assert(child->real_type == OBJ_REF_DELTA);
 810                 resolve_delta(child, base, result);
 811                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 812                         free_base_data(base);
 813
 814                 base->ref_first++;
 815                 return result;
 816         }
 817
 818         if (base->ofs_first <= base->ofs_last) {
 819                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 820                 struct base_data *result = alloc_base_data();
 821
 822                 assert(child->real_type == OBJ_OFS_DELTA);
 823                 resolve_delta(child, base, result);
 824                 if (base->ofs_first == base->ofs_last)
 825                         free_base_data(base);
 826
 827                 base->ofs_first++;
 828                 return result;
 829         }
 830
 831         unlink_base_data(base);
 832         return NULL;
 833 }
 834
 835 static void find_unresolved_deltas(struct base_data *base)
 836 {
 837         struct base_data *new_base, *prev_base = NULL;
 838         for (;;) {
 839                 new_base = find_unresolved_deltas_1(base, prev_base);
 840
 841                 if (new_base) {
 842                         prev_base = base;
 843                         base = new_base;
 844                 } else {
 845                         free(base);
 846                         base = prev_base;
 847                         if (!base)
 848                                 return;
 849                         prev_base = base->base;
 850                 }
 851         }
 852 }
 853
 854 static int compare_delta_entry(const void *a, const void *b)
 855 {
 856         const struct delta_entry *delta_a = a;
 857         const struct delta_entry *delta_b = b;
 858
 859         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 860         return compare_delta_bases(&delta_a->base, &delta_b->base,
 861                                    objects[delta_a->obj_no].type,
 862                                    objects[delta_b->obj_no].type);
 863 }
 864
 865 static void resolve_base(struct object_entry *obj)
 866 {
 867         struct base_data *base_obj = alloc_base_data();
 868         base_obj->obj = obj;
 869         base_obj->data = NULL;
 870         find_unresolved_deltas(base_obj);
 871 }
 872
 873 #ifndef NO_PTHREADS
 874 static void *threaded_second_pass(void *data)
 875 {
 876         set_thread_data(data);
 877         for (;;) {
 878                 int i;
 879                 work_lock();
 880                 display_progress(progress, nr_resolved_deltas);
 881                 while (nr_dispatched < nr_objects &&
 882                        is_delta_type(objects[nr_dispatched].type))
 883                         nr_dispatched++;
 884                 if (nr_dispatched >= nr_objects) {
 885                         work_unlock();
 886                         break;
 887                 }
 888                 i = nr_dispatched++;
 889                 work_unlock();
 890
 891                 resolve_base(&objects[i]);
 892         }
 893         return NULL;
 894 }
 895 #endif
 896
 897 /*
 898  * First pass:
 899  * - find locations of all objects;
 900  * - calculate SHA1 of all non-delta objects;
 901  * - remember base (SHA1 or offset) for all deltas.
 902  */
 903 static void parse_pack_objects(unsigned char *sha1)
 904 {
 905         int i, nr_delays = 0;
 906         struct delta_entry *delta = deltas;
 907         struct stat st;
 908
 909         if (verbose)
 910                 progress = start_progress(
 911                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
 912                                 nr_objects);
 913         for (i = 0; i < nr_objects; i++) {
 914                 struct object_entry *obj = &objects[i];
 915                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
 916                 obj->real_type = obj->type;
 917                 if (is_delta_type(obj->type)) {
 918                         nr_deltas++;
 919                         delta->obj_no = i;
 920                         delta++;
 921                 } else if (!data) {
 922                         /* large blobs, check later */
 923                         obj->real_type = OBJ_BAD;
 924                         nr_delays++;
 925                 } else
 926                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
 927                 free(data);
 928                 display_progress(progress, i+1);
 929         }
 930         objects[i].idx.offset = consumed_bytes;
 931         stop_progress(&progress);
 932
 933         /* Check pack integrity */
 934         flush();
 935         git_SHA1_Final(sha1, &input_ctx);
 936         if (hashcmp(fill(20), sha1))
 937                 die(_("pack is corrupted (SHA1 mismatch)"));
 938         use(20);
 939
 940         /* If input_fd is a file, we should have reached its end now. */
 941         if (fstat(input_fd, &st))
 942                 die_errno(_("cannot fstat packfile"));
 943         if (S_ISREG(st.st_mode) &&
 944                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 945                 die(_("pack has junk at the end"));
 946
 947         for (i = 0; i < nr_objects; i++) {
 948                 struct object_entry *obj = &objects[i];
 949                 if (obj->real_type != OBJ_BAD)
 950                         continue;
 951                 obj->real_type = obj->type;
 952                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
 953                 nr_delays--;
 954         }
 955         if (nr_delays)
 956                 die(_("confusion beyond insanity in parse_pack_objects()"));
 957 }
 958
 959 /*
 960  * Second pass:
 961  * - for all non-delta objects, look if it is used as a base for
 962  *   deltas;
 963  * - if used as a base, uncompress the object and apply all deltas,
 964  *   recursively checking if the resulting object is used as a base
 965  *   for some more deltas.
 966  */
 967 static void resolve_deltas(void)
 968 {
 969         int i;
 970
 971         if (!nr_deltas)
 972                 return;
 973
 974         /* Sort deltas by base SHA1/offset for fast searching */
 975         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 976               compare_delta_entry);
 977
 978         if (verbose)
 979                 progress = start_progress(_("Resolving deltas"), nr_deltas);
 980
 981 #ifndef NO_PTHREADS
 982         nr_dispatched = 0;
 983         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
 984                 init_thread();
 985                 for (i = 0; i < nr_threads; i++) {
 986                         int ret = pthread_create(&thread_data[i].thread, NULL,
 987                                                  threaded_second_pass, thread_data + i);
 988                         if (ret)
 989                                 die("unable to create thread: %s", strerror(ret));
 990                 }
 991                 for (i = 0; i < nr_threads; i++)
 992                         pthread_join(thread_data[i].thread, NULL);
 993                 cleanup_thread();
 994                 return;
 995         }
 996 #endif
 997
 998         for (i = 0; i < nr_objects; i++) {
 999                 struct object_entry *obj = &objects[i];
1000
1001                 if (is_delta_type(obj->type))
1002                         continue;
1003                 resolve_base(obj);
1004                 display_progress(progress, nr_resolved_deltas);
1005         }
1006 }
1007
1008 /*
1009  * Third pass:
1010  * - append objects to convert thin pack to full pack if required
1011  * - write the final 20-byte SHA-1
1012  */
1013 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1014 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1015 {
1016         if (nr_deltas == nr_resolved_deltas) {
1017                 stop_progress(&progress);
1018                 /* Flush remaining pack final 20-byte SHA1. */
1019                 flush();
1020                 return;
1021         }
1022
1023         if (fix_thin_pack) {
1024                 struct sha1file *f;
1025                 unsigned char read_sha1[20], tail_sha1[20];
1026                 char msg[48];
1027                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1028                 int nr_objects_initial = nr_objects;
1029                 if (nr_unresolved <= 0)
1030                         die(_("confusion beyond insanity"));
1031                 objects = xrealloc(objects,
1032                                    (nr_objects + nr_unresolved + 1)
1033                                    * sizeof(*objects));
1034                 f = sha1fd(output_fd, curr_pack);
1035                 fix_unresolved_deltas(f, nr_unresolved);
1036                 sprintf(msg, "completed with %d local objects",
1037                         nr_objects - nr_objects_initial);
1038                 stop_progress_msg(&progress, msg);
1039                 sha1close(f, tail_sha1, 0);
1040                 hashcpy(read_sha1, pack_sha1);
1041                 fixup_pack_header_footer(output_fd, pack_sha1,
1042                                          curr_pack, nr_objects,
1043                                          read_sha1, consumed_bytes-20);
1044                 if (hashcmp(read_sha1, tail_sha1) != 0)
1045                         die("Unexpected tail checksum for %s "
1046                             "(disk corruption?)", curr_pack);
1047         }
1048         if (nr_deltas != nr_resolved_deltas)
1049                 die(Q_("pack has %d unresolved delta",
1050                        "pack has %d unresolved deltas",
1051                        nr_deltas - nr_resolved_deltas),
1052                     nr_deltas - nr_resolved_deltas);
1053 }
1054
1055 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1056 {
1057         git_zstream stream;
1058         int status;
1059         unsigned char outbuf[4096];
1060
1061         memset(&stream, 0, sizeof(stream));
1062         git_deflate_init(&stream, zlib_compression_level);
1063         stream.next_in = in;
1064         stream.avail_in = size;
1065
1066         do {
1067                 stream.next_out = outbuf;
1068                 stream.avail_out = sizeof(outbuf);
1069                 status = git_deflate(&stream, Z_FINISH);
1070                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1071         } while (status == Z_OK);
1072
1073         if (status != Z_STREAM_END)
1074                 die(_("unable to deflate appended object (%d)"), status);
1075         size = stream.total_out;
1076         git_deflate_end(&stream);
1077         return size;
1078 }
1079
1080 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1081                                const unsigned char *sha1, void *buf,
1082                                unsigned long size, enum object_type type)
1083 {
1084         struct object_entry *obj = &objects[nr_objects++];
1085         unsigned char header[10];
1086         unsigned long s = size;
1087         int n = 0;
1088         unsigned char c = (type << 4) | (s & 15);
1089         s >>= 4;
1090         while (s) {
1091                 header[n++] = c | 0x80;
1092                 c = s & 0x7f;
1093                 s >>= 7;
1094         }
1095         header[n++] = c;
1096         crc32_begin(f);
1097         sha1write(f, header, n);
1098         obj[0].size = size;
1099         obj[0].hdr_size = n;
1100         obj[0].type = type;
1101         obj[0].real_type = type;
1102         obj[1].idx.offset = obj[0].idx.offset + n;
1103         obj[1].idx.offset += write_compressed(f, buf, size);
1104         obj[0].idx.crc32 = crc32_end(f);
1105         sha1flush(f);
1106         hashcpy(obj->idx.sha1, sha1);
1107         return obj;
1108 }
1109
1110 static int delta_pos_compare(const void *_a, const void *_b)
1111 {
1112         struct delta_entry *a = *(struct delta_entry **)_a;
1113         struct delta_entry *b = *(struct delta_entry **)_b;
1114         return a->obj_no - b->obj_no;
1115 }
1116
1117 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1118 {
1119         struct delta_entry **sorted_by_pos;
1120         int i, n = 0;
1121
1122         /*
1123          * Since many unresolved deltas may well be themselves base objects
1124          * for more unresolved deltas, we really want to include the
1125          * smallest number of base objects that would cover as much delta
1126          * as possible by picking the
1127          * trunc deltas first, allowing for other deltas to resolve without
1128          * additional base objects.  Since most base objects are to be found
1129          * before deltas depending on them, a good heuristic is to start
1130          * resolving deltas in the same order as their position in the pack.
1131          */
1132         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1133         for (i = 0; i < nr_deltas; i++) {
1134                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1135                         continue;
1136                 sorted_by_pos[n++] = &deltas[i];
1137         }
1138         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1139
1140         for (i = 0; i < n; i++) {
1141                 struct delta_entry *d = sorted_by_pos[i];
1142                 enum object_type type;
1143                 struct base_data *base_obj = alloc_base_data();
1144
1145                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1146                         continue;
1147                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1148                 if (!base_obj->data)
1149                         continue;
1150
1151                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1152                                 base_obj->size, typename(type)))
1153                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1154                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1155                                         base_obj->data, base_obj->size, type);
1156                 find_unresolved_deltas(base_obj);
1157                 display_progress(progress, nr_resolved_deltas);
1158         }
1159         free(sorted_by_pos);
1160 }
1161
1162 static void final(const char *final_pack_name, const char *curr_pack_name,
1163                   const char *final_index_name, const char *curr_index_name,
1164                   const char *keep_name, const char *keep_msg,
1165                   unsigned char *sha1)
1166 {
1167         const char *report = "pack";
1168         char name[PATH_MAX];
1169         int err;
1170
1171         if (!from_stdin) {
1172                 close(input_fd);
1173         } else {
1174                 fsync_or_die(output_fd, curr_pack_name);
1175                 err = close(output_fd);
1176                 if (err)
1177                         die_errno(_("error while closing pack file"));
1178         }
1179
1180         if (keep_msg) {
1181                 int keep_fd, keep_msg_len = strlen(keep_msg);
1182
1183                 if (!keep_name)
1184                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1185                 else
1186                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1187
1188                 if (keep_fd < 0) {
1189                         if (errno != EEXIST)
1190                                 die_errno(_("cannot write keep file '%s'"),
1191                                           keep_name);
1192                 } else {
1193                         if (keep_msg_len > 0) {
1194                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1195                                 write_or_die(keep_fd, "\n", 1);
1196                         }
1197                         if (close(keep_fd) != 0)
1198                                 die_errno(_("cannot close written keep file '%s'"),
1199                                     keep_name);
1200                         report = "keep";
1201                 }
1202         }
1203
1204         if (final_pack_name != curr_pack_name) {
1205                 if (!final_pack_name) {
1206                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1207                                  get_object_directory(), sha1_to_hex(sha1));
1208                         final_pack_name = name;
1209                 }
1210                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1211                         die(_("cannot store pack file"));
1212         } else if (from_stdin)
1213                 chmod(final_pack_name, 0444);
1214
1215         if (final_index_name != curr_index_name) {
1216                 if (!final_index_name) {
1217                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1218                                  get_object_directory(), sha1_to_hex(sha1));
1219                         final_index_name = name;
1220                 }
1221                 if (move_temp_to_file(curr_index_name, final_index_name))
1222                         die(_("cannot store index file"));
1223         } else
1224                 chmod(final_index_name, 0444);
1225
1226         if (!from_stdin) {
1227                 printf("%s\n", sha1_to_hex(sha1));
1228         } else {
1229                 char buf[48];
1230                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1231                                    report, sha1_to_hex(sha1));
1232                 write_or_die(1, buf, len);
1233
1234                 /*
1235                  * Let's just mimic git-unpack-objects here and write
1236                  * the last part of the input buffer to stdout.
1237                  */
1238                 while (input_len) {
1239                         err = xwrite(1, input_buffer + input_offset, input_len);
1240                         if (err <= 0)
1241                                 break;
1242                         input_len -= err;
1243                         input_offset += err;
1244                 }
1245         }
1246 }
1247
1248 static int git_index_pack_config(const char *k, const char *v, void *cb)
1249 {
1250         struct pack_idx_option *opts = cb;
1251
1252         if (!strcmp(k, "pack.indexversion")) {
1253                 opts->version = git_config_int(k, v);
1254                 if (opts->version > 2)
1255                         die("bad pack.indexversion=%"PRIu32, opts->version);
1256                 return 0;
1257         }
1258         if (!strcmp(k, "pack.threads")) {
1259                 nr_threads = git_config_int(k, v);
1260                 if (nr_threads < 0)
1261                         die("invalid number of threads specified (%d)",
1262                             nr_threads);
1263 #ifdef NO_PTHREADS
1264                 if (nr_threads != 1)
1265                         warning("no threads support, ignoring %s", k);
1266                 nr_threads = 1;
1267 #endif
1268                 return 0;
1269         }
1270         return git_default_config(k, v, cb);
1271 }
1272
1273 static int cmp_uint32(const void *a_, const void *b_)
1274 {
1275         uint32_t a = *((uint32_t *)a_);
1276         uint32_t b = *((uint32_t *)b_);
1277
1278         return (a < b) ? -1 : (a != b);
1279 }
1280
1281 static void read_v2_anomalous_offsets(struct packed_git *p,
1282                                       struct pack_idx_option *opts)
1283 {
1284         const uint32_t *idx1, *idx2;
1285         uint32_t i;
1286
1287         /* The address of the 4-byte offset table */
1288         idx1 = (((const uint32_t *)p->index_data)
1289                 + 2 /* 8-byte header */
1290                 + 256 /* fan out */
1291                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1292                 + p->num_objects /* CRC32 table */
1293                 );
1294
1295         /* The address of the 8-byte offset table */
1296         idx2 = idx1 + p->num_objects;
1297
1298         for (i = 0; i < p->num_objects; i++) {
1299                 uint32_t off = ntohl(idx1[i]);
1300                 if (!(off & 0x80000000))
1301                         continue;
1302                 off = off & 0x7fffffff;
1303                 if (idx2[off * 2])
1304                         continue;
1305                 /*
1306                  * The real offset is ntohl(idx2[off * 2]) in high 4
1307                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1308                  * octets.  But idx2[off * 2] is Zero!!!
1309                  */
1310                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1311                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1312         }
1313
1314         if (1 < opts->anomaly_nr)
1315                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1316 }
1317
1318 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1319 {
1320         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1321
1322         if (!p)
1323                 die(_("Cannot open existing pack file '%s'"), pack_name);
1324         if (open_pack_index(p))
1325                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1326
1327         /* Read the attributes from the existing idx file */
1328         opts->version = p->index_version;
1329
1330         if (opts->version == 2)
1331                 read_v2_anomalous_offsets(p, opts);
1332
1333         /*
1334          * Get rid of the idx file as we do not need it anymore.
1335          * NEEDSWORK: extract this bit from free_pack_by_name() in
1336          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1337          * know we haven't installed this pack (hence we never have
1338          * read anything from it).
1339          */
1340         close_pack_index(p);
1341         free(p);
1342 }
1343
1344 static void show_pack_info(int stat_only)
1345 {
1346         int i, baseobjects = nr_objects - nr_deltas;
1347         unsigned long *chain_histogram = NULL;
1348
1349         if (deepest_delta)
1350                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1351
1352         for (i = 0; i < nr_objects; i++) {
1353                 struct object_entry *obj = &objects[i];
1354
1355                 if (is_delta_type(obj->type))
1356                         chain_histogram[obj->delta_depth - 1]++;
1357                 if (stat_only)
1358                         continue;
1359                 printf("%s %-6s %lu %lu %"PRIuMAX,
1360                        sha1_to_hex(obj->idx.sha1),
1361                        typename(obj->real_type), obj->size,
1362                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1363                        (uintmax_t)obj->idx.offset);
1364                 if (is_delta_type(obj->type)) {
1365                         struct object_entry *bobj = &objects[obj->base_object_no];
1366                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1367                 }
1368                 putchar('\n');
1369         }
1370
1371         if (baseobjects)
1372                 printf_ln(Q_("non delta: %d object",
1373                              "non delta: %d objects",
1374                              baseobjects),
1375                           baseobjects);
1376         for (i = 0; i < deepest_delta; i++) {
1377                 if (!chain_histogram[i])
1378                         continue;
1379                 printf_ln(Q_("chain length = %d: %lu object",
1380                              "chain length = %d: %lu objects",
1381                              chain_histogram[i]),
1382                           i + 1,
1383                           chain_histogram[i]);
1384         }
1385 }
1386
1387 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1388 {
1389         int i, fix_thin_pack = 0, verify = 0, stat_only = 0, stat = 0;
1390         const char *curr_pack, *curr_index;
1391         const char *index_name = NULL, *pack_name = NULL;
1392         const char *keep_name = NULL, *keep_msg = NULL;
1393         char *index_name_buf = NULL, *keep_name_buf = NULL;
1394         struct pack_idx_entry **idx_objects;
1395         struct pack_idx_option opts;
1396         unsigned char pack_sha1[20];
1397
1398         if (argc == 2 && !strcmp(argv[1], "-h"))
1399                 usage(index_pack_usage);
1400
1401         read_replace_refs = 0;
1402
1403         reset_pack_idx_option(&opts);
1404         git_config(git_index_pack_config, &opts);
1405         if (prefix && chdir(prefix))
1406                 die(_("Cannot come back to cwd"));
1407
1408         for (i = 1; i < argc; i++) {
1409                 const char *arg = argv[i];
1410
1411                 if (*arg == '-') {
1412                         if (!strcmp(arg, "--stdin")) {
1413                                 from_stdin = 1;
1414                         } else if (!strcmp(arg, "--fix-thin")) {
1415                                 fix_thin_pack = 1;
1416                         } else if (!strcmp(arg, "--strict")) {
1417                                 strict = 1;
1418                         } else if (!strcmp(arg, "--verify")) {
1419                                 verify = 1;
1420                         } else if (!strcmp(arg, "--verify-stat")) {
1421                                 verify = 1;
1422                                 stat = 1;
1423                         } else if (!strcmp(arg, "--verify-stat-only")) {
1424                                 verify = 1;
1425                                 stat = 1;
1426                                 stat_only = 1;
1427                         } else if (!strcmp(arg, "--keep")) {
1428                                 keep_msg = "";
1429                         } else if (!prefixcmp(arg, "--keep=")) {
1430                                 keep_msg = arg + 7;
1431                         } else if (!prefixcmp(arg, "--threads=")) {
1432                                 char *end;
1433                                 nr_threads = strtoul(arg+10, &end, 0);
1434                                 if (!arg[10] || *end || nr_threads < 0)
1435                                         usage(index_pack_usage);
1436 #ifdef NO_PTHREADS
1437                                 if (nr_threads != 1)
1438                                         warning("no threads support, "
1439                                                 "ignoring %s", arg);
1440                                 nr_threads = 1;
1441 #endif
1442                         } else if (!prefixcmp(arg, "--pack_header=")) {
1443                                 struct pack_header *hdr;
1444                                 char *c;
1445
1446                                 hdr = (struct pack_header *)input_buffer;
1447                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1448                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1449                                 if (*c != ',')
1450                                         die(_("bad %s"), arg);
1451                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1452                                 if (*c)
1453                                         die(_("bad %s"), arg);
1454                                 input_len = sizeof(*hdr);
1455                         } else if (!strcmp(arg, "-v")) {
1456                                 verbose = 1;
1457                         } else if (!strcmp(arg, "-o")) {
1458                                 if (index_name || (i+1) >= argc)
1459                                         usage(index_pack_usage);
1460                                 index_name = argv[++i];
1461                         } else if (!prefixcmp(arg, "--index-version=")) {
1462                                 char *c;
1463                                 opts.version = strtoul(arg + 16, &c, 10);
1464                                 if (opts.version > 2)
1465                                         die(_("bad %s"), arg);
1466                                 if (*c == ',')
1467                                         opts.off32_limit = strtoul(c+1, &c, 0);
1468                                 if (*c || opts.off32_limit & 0x80000000)
1469                                         die(_("bad %s"), arg);
1470                         } else
1471                                 usage(index_pack_usage);
1472                         continue;
1473                 }
1474
1475                 if (pack_name)
1476                         usage(index_pack_usage);
1477                 pack_name = arg;
1478         }
1479
1480         if (!pack_name && !from_stdin)
1481                 usage(index_pack_usage);
1482         if (fix_thin_pack && !from_stdin)
1483                 die(_("--fix-thin cannot be used without --stdin"));
1484         if (!index_name && pack_name) {
1485                 int len = strlen(pack_name);
1486                 if (!has_extension(pack_name, ".pack"))
1487                         die(_("packfile name '%s' does not end with '.pack'"),
1488                             pack_name);
1489                 index_name_buf = xmalloc(len);
1490                 memcpy(index_name_buf, pack_name, len - 5);
1491                 strcpy(index_name_buf + len - 5, ".idx");
1492                 index_name = index_name_buf;
1493         }
1494         if (keep_msg && !keep_name && pack_name) {
1495                 int len = strlen(pack_name);
1496                 if (!has_extension(pack_name, ".pack"))
1497                         die(_("packfile name '%s' does not end with '.pack'"),
1498                             pack_name);
1499                 keep_name_buf = xmalloc(len);
1500                 memcpy(keep_name_buf, pack_name, len - 5);
1501                 strcpy(keep_name_buf + len - 5, ".keep");
1502                 keep_name = keep_name_buf;
1503         }
1504         if (verify) {
1505                 if (!index_name)
1506                         die(_("--verify with no packfile name given"));
1507                 read_idx_option(&opts, index_name);
1508                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1509         }
1510         if (strict)
1511                 opts.flags |= WRITE_IDX_STRICT;
1512
1513 #ifndef NO_PTHREADS
1514         if (!nr_threads) {
1515                 nr_threads = online_cpus();
1516                 /* An experiment showed that more threads does not mean faster */
1517                 if (nr_threads > 3)
1518                         nr_threads = 3;
1519         }
1520 #endif
1521
1522         curr_pack = open_pack_file(pack_name);
1523         parse_pack_header();
1524         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1525         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1526         parse_pack_objects(pack_sha1);
1527         resolve_deltas();
1528         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1529         free(deltas);
1530         if (strict)
1531                 check_objects();
1532
1533         if (stat)
1534                 show_pack_info(stat_only);
1535
1536         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1537         for (i = 0; i < nr_objects; i++)
1538                 idx_objects[i] = &objects[i].idx;
1539         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1540         free(idx_objects);
1541
1542         if (!verify)
1543                 final(pack_name, curr_pack,
1544                       index_name, curr_index,
1545                       keep_name, keep_msg,
1546                       pack_sha1);
1547         else
1548                 close(input_fd);
1549         free(objects);
1550         free(index_name_buf);
1551         free(keep_name_buf);
1552         if (pack_name == NULL)
1553                 free((void *) curr_pack);
1554         if (index_name == NULL)
1555                 free((void *) curr_index);
1556
1557         return 0;
1558 }