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 "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned int hdr_size;
  22         enum object_type type;
  23         enum object_type real_type;
  24         unsigned delta_depth;
  25         int base_object_no;
  26 };
  27
  28 union delta_base {
  29         unsigned char sha1[20];
  30         off_t offset;
  31 };
  32
  33 struct base_data {
  34         struct base_data *base;
  35         struct base_data *child;
  36         struct object_entry *obj;
  37         void *data;
  38         unsigned long size;
  39         int ref_first, ref_last;
  40         int ofs_first, ofs_last;
  41 };
  42
  43 #if !defined(NO_PTHREADS) && defined(NO_THREAD_SAFE_PREAD)
  44 /* pread() emulation is not thread-safe. Disable threading. */
  45 #define NO_PTHREADS
  46 #endif
  47
  48 struct thread_local {
  49 #ifndef NO_PTHREADS
  50         pthread_t thread;
  51 #endif
  52         struct base_data *base_cache;
  53         size_t base_cache_used;
  54 };
  55
  56 /*
  57  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  58  * to memcmp() only the first 20 bytes.
  59  */
  60 #define UNION_BASE_SZ   20
  61
  62 #define FLAG_LINK (1u<<20)
  63 #define FLAG_CHECKED (1u<<21)
  64
  65 struct delta_entry {
  66         union delta_base base;
  67         int obj_no;
  68 };
  69
  70 static struct object_entry *objects;
  71 static struct delta_entry *deltas;
  72 static struct thread_local nothread_data;
  73 static int nr_objects;
  74 static int nr_deltas;
  75 static int nr_resolved_deltas;
  76 static int nr_threads;
  77
  78 static int from_stdin;
  79 static int strict;
  80 static int verbose;
  81
  82 static struct progress *progress;
  83
  84 /* We always read in 4kB chunks. */
  85 static unsigned char input_buffer[4096];
  86 static unsigned int input_offset, input_len;
  87 static off_t consumed_bytes;
  88 static unsigned deepest_delta;
  89 static git_SHA_CTX input_ctx;
  90 static uint32_t input_crc32;
  91 static int input_fd, output_fd, pack_fd;
  92
  93 #ifndef NO_PTHREADS
  94
  95 static struct thread_local *thread_data;
  96 static int nr_dispatched;
  97 static int threads_active;
  98
  99 static pthread_mutex_t read_mutex;
 100 #define read_lock()             lock_mutex(&read_mutex)
 101 #define read_unlock()           unlock_mutex(&read_mutex)
 102
 103 static pthread_mutex_t counter_mutex;
 104 #define counter_lock()          lock_mutex(&counter_mutex)
 105 #define counter_unlock()        unlock_mutex(&counter_mutex)
 106
 107 static pthread_mutex_t work_mutex;
 108 #define work_lock()             lock_mutex(&work_mutex)
 109 #define work_unlock()           unlock_mutex(&work_mutex)
 110
 111 static pthread_key_t key;
 112
 113 static inline void lock_mutex(pthread_mutex_t *mutex)
 114 {
 115         if (threads_active)
 116                 pthread_mutex_lock(mutex);
 117 }
 118
 119 static inline void unlock_mutex(pthread_mutex_t *mutex)
 120 {
 121         if (threads_active)
 122                 pthread_mutex_unlock(mutex);
 123 }
 124
 125 /*
 126  * Mutex and conditional variable can't be statically-initialized on Windows.
 127  */
 128 static void init_thread(void)
 129 {
 130         init_recursive_mutex(&read_mutex);
 131         pthread_mutex_init(&counter_mutex, NULL);
 132         pthread_mutex_init(&work_mutex, NULL);
 133         pthread_key_create(&key, NULL);
 134         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 135         threads_active = 1;
 136 }
 137
 138 static void cleanup_thread(void)
 139 {
 140         if (!threads_active)
 141                 return;
 142         threads_active = 0;
 143         pthread_mutex_destroy(&read_mutex);
 144         pthread_mutex_destroy(&counter_mutex);
 145         pthread_mutex_destroy(&work_mutex);
 146         pthread_key_delete(key);
 147         free(thread_data);
 148 }
 149
 150 #else
 151
 152 #define read_lock()
 153 #define read_unlock()
 154
 155 #define counter_lock()
 156 #define counter_unlock()
 157
 158 #define work_lock()
 159 #define work_unlock()
 160
 161 #endif
 162
 163
 164 static int mark_link(struct object *obj, int type, void *data)
 165 {
 166         if (!obj)
 167                 return -1;
 168
 169         if (type != OBJ_ANY && obj->type != type)
 170                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 171
 172         obj->flags |= FLAG_LINK;
 173         return 0;
 174 }
 175
 176 /* The content of each linked object must have been checked
 177    or it must be already present in the object database */
 178 static void check_object(struct object *obj)
 179 {
 180         if (!obj)
 181                 return;
 182
 183         if (!(obj->flags & FLAG_LINK))
 184                 return;
 185
 186         if (!(obj->flags & FLAG_CHECKED)) {
 187                 unsigned long size;
 188                 int type = sha1_object_info(obj->sha1, &size);
 189                 if (type != obj->type || type <= 0)
 190                         die(_("object of unexpected type"));
 191                 obj->flags |= FLAG_CHECKED;
 192                 return;
 193         }
 194 }
 195
 196 static void check_objects(void)
 197 {
 198         unsigned i, max;
 199
 200         max = get_max_object_index();
 201         for (i = 0; i < max; i++)
 202                 check_object(get_indexed_object(i));
 203 }
 204
 205
 206 /* Discard current buffer used content. */
 207 static void flush(void)
 208 {
 209         if (input_offset) {
 210                 if (output_fd >= 0)
 211                         write_or_die(output_fd, input_buffer, input_offset);
 212                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 213                 memmove(input_buffer, input_buffer + input_offset, input_len);
 214                 input_offset = 0;
 215         }
 216 }
 217
 218 /*
 219  * Make sure at least "min" bytes are available in the buffer, and
 220  * return the pointer to the buffer.
 221  */
 222 static void *fill(int min)
 223 {
 224         if (min <= input_len)
 225                 return input_buffer + input_offset;
 226         if (min > sizeof(input_buffer))
 227                 die(Q_("cannot fill %d byte",
 228                        "cannot fill %d bytes",
 229                        min),
 230                     min);
 231         flush();
 232         do {
 233                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 234                                 sizeof(input_buffer) - input_len);
 235                 if (ret <= 0) {
 236                         if (!ret)
 237                                 die(_("early EOF"));
 238                         die_errno(_("read error on input"));
 239                 }
 240                 input_len += ret;
 241                 if (from_stdin)
 242                         display_throughput(progress, consumed_bytes + input_len);
 243         } while (input_len < min);
 244         return input_buffer;
 245 }
 246
 247 static void use(int bytes)
 248 {
 249         if (bytes > input_len)
 250                 die(_("used more bytes than were available"));
 251         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 252         input_len -= bytes;
 253         input_offset += bytes;
 254
 255         /* make sure off_t is sufficiently large not to wrap */
 256         if (signed_add_overflows(consumed_bytes, bytes))
 257                 die(_("pack too large for current definition of off_t"));
 258         consumed_bytes += bytes;
 259 }
 260
 261 static const char *open_pack_file(const char *pack_name)
 262 {
 263         if (from_stdin) {
 264                 input_fd = 0;
 265                 if (!pack_name) {
 266                         static char tmp_file[PATH_MAX];
 267                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 268                                                 "pack/tmp_pack_XXXXXX");
 269                         pack_name = xstrdup(tmp_file);
 270                 } else
 271                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 272                 if (output_fd < 0)
 273                         die_errno(_("unable to create '%s'"), pack_name);
 274                 pack_fd = output_fd;
 275         } else {
 276                 input_fd = open(pack_name, O_RDONLY);
 277                 if (input_fd < 0)
 278                         die_errno(_("cannot open packfile '%s'"), pack_name);
 279                 output_fd = -1;
 280                 pack_fd = input_fd;
 281         }
 282         git_SHA1_Init(&input_ctx);
 283         return pack_name;
 284 }
 285
 286 static void parse_pack_header(void)
 287 {
 288         struct pack_header *hdr = fill(sizeof(struct pack_header));
 289
 290         /* Header consistency check */
 291         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 292                 die(_("pack signature mismatch"));
 293         if (!pack_version_ok(hdr->hdr_version))
 294                 die("pack version %"PRIu32" unsupported",
 295                         ntohl(hdr->hdr_version));
 296
 297         nr_objects = ntohl(hdr->hdr_entries);
 298         use(sizeof(struct pack_header));
 299 }
 300
 301 static NORETURN void bad_object(unsigned long offset, const char *format,
 302                        ...) __attribute__((format (printf, 2, 3)));
 303
 304 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 305 {
 306         va_list params;
 307         char buf[1024];
 308
 309         va_start(params, format);
 310         vsnprintf(buf, sizeof(buf), format, params);
 311         va_end(params);
 312         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 313 }
 314
 315 static inline struct thread_local *get_thread_data(void)
 316 {
 317 #ifndef NO_PTHREADS
 318         if (threads_active)
 319                 return pthread_getspecific(key);
 320         assert(!threads_active &&
 321                "This should only be reached when all threads are gone");
 322 #endif
 323         return &nothread_data;
 324 }
 325
 326 #ifndef NO_PTHREADS
 327 static void set_thread_data(struct thread_local *data)
 328 {
 329         if (threads_active)
 330                 pthread_setspecific(key, data);
 331 }
 332 #endif
 333
 334 static struct base_data *alloc_base_data(void)
 335 {
 336         struct base_data *base = xmalloc(sizeof(struct base_data));
 337         memset(base, 0, sizeof(*base));
 338         base->ref_last = -1;
 339         base->ofs_last = -1;
 340         return base;
 341 }
 342
 343 static void free_base_data(struct base_data *c)
 344 {
 345         if (c->data) {
 346                 free(c->data);
 347                 c->data = NULL;
 348                 get_thread_data()->base_cache_used -= c->size;
 349         }
 350 }
 351
 352 static void prune_base_data(struct base_data *retain)
 353 {
 354         struct base_data *b;
 355         struct thread_local *data = get_thread_data();
 356         for (b = data->base_cache;
 357              data->base_cache_used > delta_base_cache_limit && b;
 358              b = b->child) {
 359                 if (b->data && b != retain)
 360                         free_base_data(b);
 361         }
 362 }
 363
 364 static void link_base_data(struct base_data *base, struct base_data *c)
 365 {
 366         if (base)
 367                 base->child = c;
 368         else
 369                 get_thread_data()->base_cache = c;
 370
 371         c->base = base;
 372         c->child = NULL;
 373         if (c->data)
 374                 get_thread_data()->base_cache_used += c->size;
 375         prune_base_data(c);
 376 }
 377
 378 static void unlink_base_data(struct base_data *c)
 379 {
 380         struct base_data *base = c->base;
 381         if (base)
 382                 base->child = NULL;
 383         else
 384                 get_thread_data()->base_cache = NULL;
 385         free_base_data(c);
 386 }
 387
 388 static int is_delta_type(enum object_type type)
 389 {
 390         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 391 }
 392
 393 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 394                                enum object_type type, unsigned char *sha1)
 395 {
 396         static char fixed_buf[8192];
 397         int status;
 398         git_zstream stream;
 399         void *buf;
 400         git_SHA_CTX c;
 401         char hdr[32];
 402         int hdrlen;
 403
 404         if (!is_delta_type(type)) {
 405                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 406                 git_SHA1_Init(&c);
 407                 git_SHA1_Update(&c, hdr, hdrlen);
 408         } else
 409                 sha1 = NULL;
 410         if (type == OBJ_BLOB && size > big_file_threshold)
 411                 buf = fixed_buf;
 412         else
 413                 buf = xmalloc(size);
 414
 415         memset(&stream, 0, sizeof(stream));
 416         git_inflate_init(&stream);
 417         stream.next_out = buf;
 418         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 419
 420         do {
 421                 unsigned char *last_out = stream.next_out;
 422                 stream.next_in = fill(1);
 423                 stream.avail_in = input_len;
 424                 status = git_inflate(&stream, 0);
 425                 use(input_len - stream.avail_in);
 426                 if (sha1)
 427                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 428                 if (buf == fixed_buf) {
 429                         stream.next_out = buf;
 430                         stream.avail_out = sizeof(fixed_buf);
 431                 }
 432         } while (status == Z_OK);
 433         if (stream.total_out != size || status != Z_STREAM_END)
 434                 bad_object(offset, _("inflate returned %d"), status);
 435         git_inflate_end(&stream);
 436         if (sha1)
 437                 git_SHA1_Final(sha1, &c);
 438         return buf == fixed_buf ? NULL : buf;
 439 }
 440
 441 static void *unpack_raw_entry(struct object_entry *obj,
 442                               union delta_base *delta_base,
 443                               unsigned char *sha1)
 444 {
 445         unsigned char *p;
 446         unsigned long size, c;
 447         off_t base_offset;
 448         unsigned shift;
 449         void *data;
 450
 451         obj->idx.offset = consumed_bytes;
 452         input_crc32 = crc32(0, NULL, 0);
 453
 454         p = fill(1);
 455         c = *p;
 456         use(1);
 457         obj->type = (c >> 4) & 7;
 458         size = (c & 15);
 459         shift = 4;
 460         while (c & 0x80) {
 461                 p = fill(1);
 462                 c = *p;
 463                 use(1);
 464                 size += (c & 0x7f) << shift;
 465                 shift += 7;
 466         }
 467         obj->size = size;
 468
 469         switch (obj->type) {
 470         case OBJ_REF_DELTA:
 471                 hashcpy(delta_base->sha1, fill(20));
 472                 use(20);
 473                 break;
 474         case OBJ_OFS_DELTA:
 475                 memset(delta_base, 0, sizeof(*delta_base));
 476                 p = fill(1);
 477                 c = *p;
 478                 use(1);
 479                 base_offset = c & 127;
 480                 while (c & 128) {
 481                         base_offset += 1;
 482                         if (!base_offset || MSB(base_offset, 7))
 483                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 484                         p = fill(1);
 485                         c = *p;
 486                         use(1);
 487                         base_offset = (base_offset << 7) + (c & 127);
 488                 }
 489                 delta_base->offset = obj->idx.offset - base_offset;
 490                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 491                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 492                 break;
 493         case OBJ_COMMIT:
 494         case OBJ_TREE:
 495         case OBJ_BLOB:
 496         case OBJ_TAG:
 497                 break;
 498         default:
 499                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 500         }
 501         obj->hdr_size = consumed_bytes - obj->idx.offset;
 502
 503         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 504         obj->idx.crc32 = input_crc32;
 505         return data;
 506 }
 507
 508 static void *unpack_data(struct object_entry *obj,
 509                          int (*consume)(const unsigned char *, unsigned long, void *),
 510                          void *cb_data)
 511 {
 512         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 513         unsigned long len = obj[1].idx.offset - from;
 514         unsigned char *data, *inbuf;
 515         git_zstream stream;
 516         int status;
 517
 518         data = xmalloc(consume ? 64*1024 : obj->size);
 519         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 520
 521         memset(&stream, 0, sizeof(stream));
 522         git_inflate_init(&stream);
 523         stream.next_out = data;
 524         stream.avail_out = consume ? 64*1024 : obj->size;
 525
 526         do {
 527                 unsigned char *last_out = stream.next_out;
 528                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 529                 n = pread(pack_fd, inbuf, n, from);
 530                 if (n < 0)
 531                         die_errno(_("cannot pread pack file"));
 532                 if (!n)
 533                         die(Q_("premature end of pack file, %lu byte missing",
 534                                "premature end of pack file, %lu bytes missing",
 535                                len),
 536                             len);
 537                 from += n;
 538                 len -= n;
 539                 stream.next_in = inbuf;
 540                 stream.avail_in = n;
 541                 status = git_inflate(&stream, 0);
 542                 if (consume) {
 543                         if (consume(last_out, stream.next_out - last_out, cb_data)) {
 544                                 free(inbuf);
 545                                 free(data);
 546                                 return NULL;
 547                         }
 548                         stream.next_out = data;
 549                         stream.avail_out = 64*1024;
 550                 }
 551         } while (len && status == Z_OK && !stream.avail_in);
 552
 553         /* This has been inflated OK when first encountered, so... */
 554         if (status != Z_STREAM_END || stream.total_out != obj->size)
 555                 die(_("serious inflate inconsistency"));
 556
 557         git_inflate_end(&stream);
 558         free(inbuf);
 559         if (consume) {
 560                 free(data);
 561                 data = NULL;
 562         }
 563         return data;
 564 }
 565
 566 static void *get_data_from_pack(struct object_entry *obj)
 567 {
 568         return unpack_data(obj, NULL, NULL);
 569 }
 570
 571 static int compare_delta_bases(const union delta_base *base1,
 572                                const union delta_base *base2,
 573                                enum object_type type1,
 574                                enum object_type type2)
 575 {
 576         int cmp = type1 - type2;
 577         if (cmp)
 578                 return cmp;
 579         return memcmp(base1, base2, UNION_BASE_SZ);
 580 }
 581
 582 static int find_delta(const union delta_base *base, enum object_type type)
 583 {
 584         int first = 0, last = nr_deltas;
 585
 586         while (first < last) {
 587                 int next = (first + last) / 2;
 588                 struct delta_entry *delta = &deltas[next];
 589                 int cmp;
 590
 591                 cmp = compare_delta_bases(base, &delta->base,
 592                                           type, objects[delta->obj_no].type);
 593                 if (!cmp)
 594                         return next;
 595                 if (cmp < 0) {
 596                         last = next;
 597                         continue;
 598                 }
 599                 first = next+1;
 600         }
 601         return -first-1;
 602 }
 603
 604 static void find_delta_children(const union delta_base *base,
 605                                 int *first_index, int *last_index,
 606                                 enum object_type type)
 607 {
 608         int first = find_delta(base, type);
 609         int last = first;
 610         int end = nr_deltas - 1;
 611
 612         if (first < 0) {
 613                 *first_index = 0;
 614                 *last_index = -1;
 615                 return;
 616         }
 617         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 618                 --first;
 619         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 620                 ++last;
 621         *first_index = first;
 622         *last_index = last;
 623 }
 624
 625 struct compare_data {
 626         struct object_entry *entry;
 627         struct git_istream *st;
 628         unsigned char *buf;
 629         unsigned long buf_size;
 630 };
 631
 632 static int compare_objects(const unsigned char *buf, unsigned long size,
 633                            void *cb_data)
 634 {
 635         struct compare_data *data = cb_data;
 636
 637         if (data->buf_size < size) {
 638                 free(data->buf);
 639                 data->buf = xmalloc(size);
 640                 data->buf_size = size;
 641         }
 642
 643         while (size) {
 644                 ssize_t len = read_istream(data->st, data->buf, size);
 645                 if (len == 0)
 646                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 647                             sha1_to_hex(data->entry->idx.sha1));
 648                 if (len < 0)
 649                         die(_("unable to read %s"),
 650                             sha1_to_hex(data->entry->idx.sha1));
 651                 if (memcmp(buf, data->buf, len))
 652                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 653                             sha1_to_hex(data->entry->idx.sha1));
 654                 size -= len;
 655                 buf += len;
 656         }
 657         return 0;
 658 }
 659
 660 static int check_collison(struct object_entry *entry)
 661 {
 662         struct compare_data data;
 663         enum object_type type;
 664         unsigned long size;
 665
 666         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 667                 return -1;
 668
 669         memset(&data, 0, sizeof(data));
 670         data.entry = entry;
 671         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 672         if (!data.st)
 673                 return -1;
 674         if (size != entry->size || type != entry->type)
 675                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 676                     sha1_to_hex(entry->idx.sha1));
 677         unpack_data(entry, compare_objects, &data);
 678         close_istream(data.st);
 679         free(data.buf);
 680         return 0;
 681 }
 682
 683 static void sha1_object(const void *data, struct object_entry *obj_entry,
 684                         unsigned long size, enum object_type type,
 685                         const unsigned char *sha1)
 686 {
 687         void *new_data = NULL;
 688         int collision_test_needed;
 689
 690         assert(data || obj_entry);
 691
 692         read_lock();
 693         collision_test_needed = has_sha1_file(sha1);
 694         read_unlock();
 695
 696         if (collision_test_needed && !data) {
 697                 read_lock();
 698                 if (!check_collison(obj_entry))
 699                         collision_test_needed = 0;
 700                 read_unlock();
 701         }
 702         if (collision_test_needed) {
 703                 void *has_data;
 704                 enum object_type has_type;
 705                 unsigned long has_size;
 706                 read_lock();
 707                 has_type = sha1_object_info(sha1, &has_size);
 708                 if (has_type != type || has_size != size)
 709                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 710                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 711                 read_unlock();
 712                 if (!data)
 713                         data = new_data = get_data_from_pack(obj_entry);
 714                 if (!has_data)
 715                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 716                 if (size != has_size || type != has_type ||
 717                     memcmp(data, has_data, size) != 0)
 718                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 719                 free(has_data);
 720         }
 721
 722         if (strict) {
 723                 read_lock();
 724                 if (type == OBJ_BLOB) {
 725                         struct blob *blob = lookup_blob(sha1);
 726                         if (blob)
 727                                 blob->object.flags |= FLAG_CHECKED;
 728                         else
 729                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 730                 } else {
 731                         struct object *obj;
 732                         int eaten;
 733                         void *buf = (void *) data;
 734
 735                         if (!buf)
 736                                 buf = new_data = get_data_from_pack(obj_entry);
 737
 738                         /*
 739                          * we do not need to free the memory here, as the
 740                          * buf is deleted by the caller.
 741                          */
 742                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 743                         if (!obj)
 744                                 die(_("invalid %s"), typename(type));
 745                         if (fsck_object(obj, 1, fsck_error_function))
 746                                 die(_("Error in object"));
 747                         if (fsck_walk(obj, mark_link, NULL))
 748                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 749
 750                         if (obj->type == OBJ_TREE) {
 751                                 struct tree *item = (struct tree *) obj;
 752                                 item->buffer = NULL;
 753                         }
 754                         if (obj->type == OBJ_COMMIT) {
 755                                 struct commit *commit = (struct commit *) obj;
 756                                 commit->buffer = NULL;
 757                         }
 758                         obj->flags |= FLAG_CHECKED;
 759                 }
 760                 read_unlock();
 761         }
 762
 763         free(new_data);
 764 }
 765
 766 /*
 767  * This function is part of find_unresolved_deltas(). There are two
 768  * walkers going in the opposite ways.
 769  *
 770  * The first one in find_unresolved_deltas() traverses down from
 771  * parent node to children, deflating nodes along the way. However,
 772  * memory for deflated nodes is limited by delta_base_cache_limit, so
 773  * at some point parent node's deflated content may be freed.
 774  *
 775  * The second walker is this function, which goes from current node up
 776  * to top parent if necessary to deflate the node. In normal
 777  * situation, its parent node would be already deflated, so it just
 778  * needs to apply delta.
 779  *
 780  * In the worst case scenario, parent node is no longer deflated because
 781  * we're running out of delta_base_cache_limit; we need to re-deflate
 782  * parents, possibly up to the top base.
 783  *
 784  * All deflated objects here are subject to be freed if we exceed
 785  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 786  * just need to make sure the last node is not freed.
 787  */
 788 static void *get_base_data(struct base_data *c)
 789 {
 790         if (!c->data) {
 791                 struct object_entry *obj = c->obj;
 792                 struct base_data **delta = NULL;
 793                 int delta_nr = 0, delta_alloc = 0;
 794
 795                 while (is_delta_type(c->obj->type) && !c->data) {
 796                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 797                         delta[delta_nr++] = c;
 798                         c = c->base;
 799                 }
 800                 if (!delta_nr) {
 801                         c->data = get_data_from_pack(obj);
 802                         c->size = obj->size;
 803                         get_thread_data()->base_cache_used += c->size;
 804                         prune_base_data(c);
 805                 }
 806                 for (; delta_nr > 0; delta_nr--) {
 807                         void *base, *raw;
 808                         c = delta[delta_nr - 1];
 809                         obj = c->obj;
 810                         base = get_base_data(c->base);
 811                         raw = get_data_from_pack(obj);
 812                         c->data = patch_delta(
 813                                 base, c->base->size,
 814                                 raw, obj->size,
 815                                 &c->size);
 816                         free(raw);
 817                         if (!c->data)
 818                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 819                         get_thread_data()->base_cache_used += c->size;
 820                         prune_base_data(c);
 821                 }
 822                 free(delta);
 823         }
 824         return c->data;
 825 }
 826
 827 static void resolve_delta(struct object_entry *delta_obj,
 828                           struct base_data *base, struct base_data *result)
 829 {
 830         void *base_data, *delta_data;
 831
 832         delta_obj->real_type = base->obj->real_type;
 833         delta_obj->delta_depth = base->obj->delta_depth + 1;
 834         if (deepest_delta < delta_obj->delta_depth)
 835                 deepest_delta = delta_obj->delta_depth;
 836         delta_obj->base_object_no = base->obj - objects;
 837         delta_data = get_data_from_pack(delta_obj);
 838         base_data = get_base_data(base);
 839         result->obj = delta_obj;
 840         result->data = patch_delta(base_data, base->size,
 841                                    delta_data, delta_obj->size, &result->size);
 842         free(delta_data);
 843         if (!result->data)
 844                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 845         hash_sha1_file(result->data, result->size,
 846                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 847         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 848                     delta_obj->idx.sha1);
 849         counter_lock();
 850         nr_resolved_deltas++;
 851         counter_unlock();
 852 }
 853
 854 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 855                                                   struct base_data *prev_base)
 856 {
 857         if (base->ref_last == -1 && base->ofs_last == -1) {
 858                 union delta_base base_spec;
 859
 860                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 861                 find_delta_children(&base_spec,
 862                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 863
 864                 memset(&base_spec, 0, sizeof(base_spec));
 865                 base_spec.offset = base->obj->idx.offset;
 866                 find_delta_children(&base_spec,
 867                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 868
 869                 if (base->ref_last == -1 && base->ofs_last == -1) {
 870                         free(base->data);
 871                         return NULL;
 872                 }
 873
 874                 link_base_data(prev_base, base);
 875         }
 876
 877         if (base->ref_first <= base->ref_last) {
 878                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 879                 struct base_data *result = alloc_base_data();
 880
 881                 assert(child->real_type == OBJ_REF_DELTA);
 882                 resolve_delta(child, base, result);
 883                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 884                         free_base_data(base);
 885
 886                 base->ref_first++;
 887                 return result;
 888         }
 889
 890         if (base->ofs_first <= base->ofs_last) {
 891                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 892                 struct base_data *result = alloc_base_data();
 893
 894                 assert(child->real_type == OBJ_OFS_DELTA);
 895                 resolve_delta(child, base, result);
 896                 if (base->ofs_first == base->ofs_last)
 897                         free_base_data(base);
 898
 899                 base->ofs_first++;
 900                 return result;
 901         }
 902
 903         unlink_base_data(base);
 904         return NULL;
 905 }
 906
 907 static void find_unresolved_deltas(struct base_data *base)
 908 {
 909         struct base_data *new_base, *prev_base = NULL;
 910         for (;;) {
 911                 new_base = find_unresolved_deltas_1(base, prev_base);
 912
 913                 if (new_base) {
 914                         prev_base = base;
 915                         base = new_base;
 916                 } else {
 917                         free(base);
 918                         base = prev_base;
 919                         if (!base)
 920                                 return;
 921                         prev_base = base->base;
 922                 }
 923         }
 924 }
 925
 926 static int compare_delta_entry(const void *a, const void *b)
 927 {
 928         const struct delta_entry *delta_a = a;
 929         const struct delta_entry *delta_b = b;
 930
 931         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 932         return compare_delta_bases(&delta_a->base, &delta_b->base,
 933                                    objects[delta_a->obj_no].type,
 934                                    objects[delta_b->obj_no].type);
 935 }
 936
 937 static void resolve_base(struct object_entry *obj)
 938 {
 939         struct base_data *base_obj = alloc_base_data();
 940         base_obj->obj = obj;
 941         base_obj->data = NULL;
 942         find_unresolved_deltas(base_obj);
 943 }
 944
 945 #ifndef NO_PTHREADS
 946 static void *threaded_second_pass(void *data)
 947 {
 948         set_thread_data(data);
 949         for (;;) {
 950                 int i;
 951                 work_lock();
 952                 display_progress(progress, nr_resolved_deltas);
 953                 while (nr_dispatched < nr_objects &&
 954                        is_delta_type(objects[nr_dispatched].type))
 955                         nr_dispatched++;
 956                 if (nr_dispatched >= nr_objects) {
 957                         work_unlock();
 958                         break;
 959                 }
 960                 i = nr_dispatched++;
 961                 work_unlock();
 962
 963                 resolve_base(&objects[i]);
 964         }
 965         return NULL;
 966 }
 967 #endif
 968
 969 /*
 970  * First pass:
 971  * - find locations of all objects;
 972  * - calculate SHA1 of all non-delta objects;
 973  * - remember base (SHA1 or offset) for all deltas.
 974  */
 975 static void parse_pack_objects(unsigned char *sha1)
 976 {
 977         int i, nr_delays = 0;
 978         struct delta_entry *delta = deltas;
 979         struct stat st;
 980
 981         if (verbose)
 982                 progress = start_progress(
 983                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
 984                                 nr_objects);
 985         for (i = 0; i < nr_objects; i++) {
 986                 struct object_entry *obj = &objects[i];
 987                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
 988                 obj->real_type = obj->type;
 989                 if (is_delta_type(obj->type)) {
 990                         nr_deltas++;
 991                         delta->obj_no = i;
 992                         delta++;
 993                 } else if (!data) {
 994                         /* large blobs, check later */
 995                         obj->real_type = OBJ_BAD;
 996                         nr_delays++;
 997                 } else
 998                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
 999                 free(data);
1000                 display_progress(progress, i+1);
1001         }
1002         objects[i].idx.offset = consumed_bytes;
1003         stop_progress(&progress);
1004
1005         /* Check pack integrity */
1006         flush();
1007         git_SHA1_Final(sha1, &input_ctx);
1008         if (hashcmp(fill(20), sha1))
1009                 die(_("pack is corrupted (SHA1 mismatch)"));
1010         use(20);
1011
1012         /* If input_fd is a file, we should have reached its end now. */
1013         if (fstat(input_fd, &st))
1014                 die_errno(_("cannot fstat packfile"));
1015         if (S_ISREG(st.st_mode) &&
1016                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1017                 die(_("pack has junk at the end"));
1018
1019         for (i = 0; i < nr_objects; i++) {
1020                 struct object_entry *obj = &objects[i];
1021                 if (obj->real_type != OBJ_BAD)
1022                         continue;
1023                 obj->real_type = obj->type;
1024                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1025                 nr_delays--;
1026         }
1027         if (nr_delays)
1028                 die(_("confusion beyond insanity in parse_pack_objects()"));
1029 }
1030
1031 /*
1032  * Second pass:
1033  * - for all non-delta objects, look if it is used as a base for
1034  *   deltas;
1035  * - if used as a base, uncompress the object and apply all deltas,
1036  *   recursively checking if the resulting object is used as a base
1037  *   for some more deltas.
1038  */
1039 static void resolve_deltas(void)
1040 {
1041         int i;
1042
1043         if (!nr_deltas)
1044                 return;
1045
1046         /* Sort deltas by base SHA1/offset for fast searching */
1047         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1048               compare_delta_entry);
1049
1050         if (verbose)
1051                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1052
1053 #ifndef NO_PTHREADS
1054         nr_dispatched = 0;
1055         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1056                 init_thread();
1057                 for (i = 0; i < nr_threads; i++) {
1058                         int ret = pthread_create(&thread_data[i].thread, NULL,
1059                                                  threaded_second_pass, thread_data + i);
1060                         if (ret)
1061                                 die("unable to create thread: %s", strerror(ret));
1062                 }
1063                 for (i = 0; i < nr_threads; i++)
1064                         pthread_join(thread_data[i].thread, NULL);
1065                 cleanup_thread();
1066                 return;
1067         }
1068 #endif
1069
1070         for (i = 0; i < nr_objects; i++) {
1071                 struct object_entry *obj = &objects[i];
1072
1073                 if (is_delta_type(obj->type))
1074                         continue;
1075                 resolve_base(obj);
1076                 display_progress(progress, nr_resolved_deltas);
1077         }
1078 }
1079
1080 /*
1081  * Third pass:
1082  * - append objects to convert thin pack to full pack if required
1083  * - write the final 20-byte SHA-1
1084  */
1085 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1086 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1087 {
1088         if (nr_deltas == nr_resolved_deltas) {
1089                 stop_progress(&progress);
1090                 /* Flush remaining pack final 20-byte SHA1. */
1091                 flush();
1092                 return;
1093         }
1094
1095         if (fix_thin_pack) {
1096                 struct sha1file *f;
1097                 unsigned char read_sha1[20], tail_sha1[20];
1098                 char msg[48];
1099                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1100                 int nr_objects_initial = nr_objects;
1101                 if (nr_unresolved <= 0)
1102                         die(_("confusion beyond insanity"));
1103                 objects = xrealloc(objects,
1104                                    (nr_objects + nr_unresolved + 1)
1105                                    * sizeof(*objects));
1106                 f = sha1fd(output_fd, curr_pack);
1107                 fix_unresolved_deltas(f, nr_unresolved);
1108                 sprintf(msg, "completed with %d local objects",
1109                         nr_objects - nr_objects_initial);
1110                 stop_progress_msg(&progress, msg);
1111                 sha1close(f, tail_sha1, 0);
1112                 hashcpy(read_sha1, pack_sha1);
1113                 fixup_pack_header_footer(output_fd, pack_sha1,
1114                                          curr_pack, nr_objects,
1115                                          read_sha1, consumed_bytes-20);
1116                 if (hashcmp(read_sha1, tail_sha1) != 0)
1117                         die("Unexpected tail checksum for %s "
1118                             "(disk corruption?)", curr_pack);
1119         }
1120         if (nr_deltas != nr_resolved_deltas)
1121                 die(Q_("pack has %d unresolved delta",
1122                        "pack has %d unresolved deltas",
1123                        nr_deltas - nr_resolved_deltas),
1124                     nr_deltas - nr_resolved_deltas);
1125 }
1126
1127 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1128 {
1129         git_zstream stream;
1130         int status;
1131         unsigned char outbuf[4096];
1132
1133         memset(&stream, 0, sizeof(stream));
1134         git_deflate_init(&stream, zlib_compression_level);
1135         stream.next_in = in;
1136         stream.avail_in = size;
1137
1138         do {
1139                 stream.next_out = outbuf;
1140                 stream.avail_out = sizeof(outbuf);
1141                 status = git_deflate(&stream, Z_FINISH);
1142                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1143         } while (status == Z_OK);
1144
1145         if (status != Z_STREAM_END)
1146                 die(_("unable to deflate appended object (%d)"), status);
1147         size = stream.total_out;
1148         git_deflate_end(&stream);
1149         return size;
1150 }
1151
1152 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1153                                const unsigned char *sha1, void *buf,
1154                                unsigned long size, enum object_type type)
1155 {
1156         struct object_entry *obj = &objects[nr_objects++];
1157         unsigned char header[10];
1158         unsigned long s = size;
1159         int n = 0;
1160         unsigned char c = (type << 4) | (s & 15);
1161         s >>= 4;
1162         while (s) {
1163                 header[n++] = c | 0x80;
1164                 c = s & 0x7f;
1165                 s >>= 7;
1166         }
1167         header[n++] = c;
1168         crc32_begin(f);
1169         sha1write(f, header, n);
1170         obj[0].size = size;
1171         obj[0].hdr_size = n;
1172         obj[0].type = type;
1173         obj[0].real_type = type;
1174         obj[1].idx.offset = obj[0].idx.offset + n;
1175         obj[1].idx.offset += write_compressed(f, buf, size);
1176         obj[0].idx.crc32 = crc32_end(f);
1177         sha1flush(f);
1178         hashcpy(obj->idx.sha1, sha1);
1179         return obj;
1180 }
1181
1182 static int delta_pos_compare(const void *_a, const void *_b)
1183 {
1184         struct delta_entry *a = *(struct delta_entry **)_a;
1185         struct delta_entry *b = *(struct delta_entry **)_b;
1186         return a->obj_no - b->obj_no;
1187 }
1188
1189 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1190 {
1191         struct delta_entry **sorted_by_pos;
1192         int i, n = 0;
1193
1194         /*
1195          * Since many unresolved deltas may well be themselves base objects
1196          * for more unresolved deltas, we really want to include the
1197          * smallest number of base objects that would cover as much delta
1198          * as possible by picking the
1199          * trunc deltas first, allowing for other deltas to resolve without
1200          * additional base objects.  Since most base objects are to be found
1201          * before deltas depending on them, a good heuristic is to start
1202          * resolving deltas in the same order as their position in the pack.
1203          */
1204         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1205         for (i = 0; i < nr_deltas; i++) {
1206                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1207                         continue;
1208                 sorted_by_pos[n++] = &deltas[i];
1209         }
1210         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1211
1212         for (i = 0; i < n; i++) {
1213                 struct delta_entry *d = sorted_by_pos[i];
1214                 enum object_type type;
1215                 struct base_data *base_obj = alloc_base_data();
1216
1217                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1218                         continue;
1219                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1220                 if (!base_obj->data)
1221                         continue;
1222
1223                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1224                                 base_obj->size, typename(type)))
1225                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1226                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1227                                         base_obj->data, base_obj->size, type);
1228                 find_unresolved_deltas(base_obj);
1229                 display_progress(progress, nr_resolved_deltas);
1230         }
1231         free(sorted_by_pos);
1232 }
1233
1234 static void final(const char *final_pack_name, const char *curr_pack_name,
1235                   const char *final_index_name, const char *curr_index_name,
1236                   const char *keep_name, const char *keep_msg,
1237                   unsigned char *sha1)
1238 {
1239         const char *report = "pack";
1240         char name[PATH_MAX];
1241         int err;
1242
1243         if (!from_stdin) {
1244                 close(input_fd);
1245         } else {
1246                 fsync_or_die(output_fd, curr_pack_name);
1247                 err = close(output_fd);
1248                 if (err)
1249                         die_errno(_("error while closing pack file"));
1250         }
1251
1252         if (keep_msg) {
1253                 int keep_fd, keep_msg_len = strlen(keep_msg);
1254
1255                 if (!keep_name)
1256                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1257                 else
1258                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1259
1260                 if (keep_fd < 0) {
1261                         if (errno != EEXIST)
1262                                 die_errno(_("cannot write keep file '%s'"),
1263                                           keep_name);
1264                 } else {
1265                         if (keep_msg_len > 0) {
1266                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1267                                 write_or_die(keep_fd, "\n", 1);
1268                         }
1269                         if (close(keep_fd) != 0)
1270                                 die_errno(_("cannot close written keep file '%s'"),
1271                                     keep_name);
1272                         report = "keep";
1273                 }
1274         }
1275
1276         if (final_pack_name != curr_pack_name) {
1277                 if (!final_pack_name) {
1278                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1279                                  get_object_directory(), sha1_to_hex(sha1));
1280                         final_pack_name = name;
1281                 }
1282                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1283                         die(_("cannot store pack file"));
1284         } else if (from_stdin)
1285                 chmod(final_pack_name, 0444);
1286
1287         if (final_index_name != curr_index_name) {
1288                 if (!final_index_name) {
1289                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1290                                  get_object_directory(), sha1_to_hex(sha1));
1291                         final_index_name = name;
1292                 }
1293                 if (move_temp_to_file(curr_index_name, final_index_name))
1294                         die(_("cannot store index file"));
1295         } else
1296                 chmod(final_index_name, 0444);
1297
1298         if (!from_stdin) {
1299                 printf("%s\n", sha1_to_hex(sha1));
1300         } else {
1301                 char buf[48];
1302                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1303                                    report, sha1_to_hex(sha1));
1304                 write_or_die(1, buf, len);
1305
1306                 /*
1307                  * Let's just mimic git-unpack-objects here and write
1308                  * the last part of the input buffer to stdout.
1309                  */
1310                 while (input_len) {
1311                         err = xwrite(1, input_buffer + input_offset, input_len);
1312                         if (err <= 0)
1313                                 break;
1314                         input_len -= err;
1315                         input_offset += err;
1316                 }
1317         }
1318 }
1319
1320 static int git_index_pack_config(const char *k, const char *v, void *cb)
1321 {
1322         struct pack_idx_option *opts = cb;
1323
1324         if (!strcmp(k, "pack.indexversion")) {
1325                 opts->version = git_config_int(k, v);
1326                 if (opts->version > 2)
1327                         die("bad pack.indexversion=%"PRIu32, opts->version);
1328                 return 0;
1329         }
1330         if (!strcmp(k, "pack.threads")) {
1331                 nr_threads = git_config_int(k, v);
1332                 if (nr_threads < 0)
1333                         die("invalid number of threads specified (%d)",
1334                             nr_threads);
1335 #ifdef NO_PTHREADS
1336                 if (nr_threads != 1)
1337                         warning("no threads support, ignoring %s", k);
1338                 nr_threads = 1;
1339 #endif
1340                 return 0;
1341         }
1342         return git_default_config(k, v, cb);
1343 }
1344
1345 static int cmp_uint32(const void *a_, const void *b_)
1346 {
1347         uint32_t a = *((uint32_t *)a_);
1348         uint32_t b = *((uint32_t *)b_);
1349
1350         return (a < b) ? -1 : (a != b);
1351 }
1352
1353 static void read_v2_anomalous_offsets(struct packed_git *p,
1354                                       struct pack_idx_option *opts)
1355 {
1356         const uint32_t *idx1, *idx2;
1357         uint32_t i;
1358
1359         /* The address of the 4-byte offset table */
1360         idx1 = (((const uint32_t *)p->index_data)
1361                 + 2 /* 8-byte header */
1362                 + 256 /* fan out */
1363                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1364                 + p->num_objects /* CRC32 table */
1365                 );
1366
1367         /* The address of the 8-byte offset table */
1368         idx2 = idx1 + p->num_objects;
1369
1370         for (i = 0; i < p->num_objects; i++) {
1371                 uint32_t off = ntohl(idx1[i]);
1372                 if (!(off & 0x80000000))
1373                         continue;
1374                 off = off & 0x7fffffff;
1375                 if (idx2[off * 2])
1376                         continue;
1377                 /*
1378                  * The real offset is ntohl(idx2[off * 2]) in high 4
1379                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1380                  * octets.  But idx2[off * 2] is Zero!!!
1381                  */
1382                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1383                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1384         }
1385
1386         if (1 < opts->anomaly_nr)
1387                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1388 }
1389
1390 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1391 {
1392         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1393
1394         if (!p)
1395                 die(_("Cannot open existing pack file '%s'"), pack_name);
1396         if (open_pack_index(p))
1397                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1398
1399         /* Read the attributes from the existing idx file */
1400         opts->version = p->index_version;
1401
1402         if (opts->version == 2)
1403                 read_v2_anomalous_offsets(p, opts);
1404
1405         /*
1406          * Get rid of the idx file as we do not need it anymore.
1407          * NEEDSWORK: extract this bit from free_pack_by_name() in
1408          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1409          * know we haven't installed this pack (hence we never have
1410          * read anything from it).
1411          */
1412         close_pack_index(p);
1413         free(p);
1414 }
1415
1416 static void show_pack_info(int stat_only)
1417 {
1418         int i, baseobjects = nr_objects - nr_deltas;
1419         unsigned long *chain_histogram = NULL;
1420
1421         if (deepest_delta)
1422                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1423
1424         for (i = 0; i < nr_objects; i++) {
1425                 struct object_entry *obj = &objects[i];
1426
1427                 if (is_delta_type(obj->type))
1428                         chain_histogram[obj->delta_depth - 1]++;
1429                 if (stat_only)
1430                         continue;
1431                 printf("%s %-6s %lu %lu %"PRIuMAX,
1432                        sha1_to_hex(obj->idx.sha1),
1433                        typename(obj->real_type), obj->size,
1434                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1435                        (uintmax_t)obj->idx.offset);
1436                 if (is_delta_type(obj->type)) {
1437                         struct object_entry *bobj = &objects[obj->base_object_no];
1438                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1439                 }
1440                 putchar('\n');
1441         }
1442
1443         if (baseobjects)
1444                 printf_ln(Q_("non delta: %d object",
1445                              "non delta: %d objects",
1446                              baseobjects),
1447                           baseobjects);
1448         for (i = 0; i < deepest_delta; i++) {
1449                 if (!chain_histogram[i])
1450                         continue;
1451                 printf_ln(Q_("chain length = %d: %lu object",
1452                              "chain length = %d: %lu objects",
1453                              chain_histogram[i]),
1454                           i + 1,
1455                           chain_histogram[i]);
1456         }
1457 }
1458
1459 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1460 {
1461         int i, fix_thin_pack = 0, verify = 0, stat_only = 0, stat = 0;
1462         const char *curr_pack, *curr_index;
1463         const char *index_name = NULL, *pack_name = NULL;
1464         const char *keep_name = NULL, *keep_msg = NULL;
1465         char *index_name_buf = NULL, *keep_name_buf = NULL;
1466         struct pack_idx_entry **idx_objects;
1467         struct pack_idx_option opts;
1468         unsigned char pack_sha1[20];
1469
1470         if (argc == 2 && !strcmp(argv[1], "-h"))
1471                 usage(index_pack_usage);
1472
1473         read_replace_refs = 0;
1474
1475         reset_pack_idx_option(&opts);
1476         git_config(git_index_pack_config, &opts);
1477         if (prefix && chdir(prefix))
1478                 die(_("Cannot come back to cwd"));
1479
1480         for (i = 1; i < argc; i++) {
1481                 const char *arg = argv[i];
1482
1483                 if (*arg == '-') {
1484                         if (!strcmp(arg, "--stdin")) {
1485                                 from_stdin = 1;
1486                         } else if (!strcmp(arg, "--fix-thin")) {
1487                                 fix_thin_pack = 1;
1488                         } else if (!strcmp(arg, "--strict")) {
1489                                 strict = 1;
1490                         } else if (!strcmp(arg, "--verify")) {
1491                                 verify = 1;
1492                         } else if (!strcmp(arg, "--verify-stat")) {
1493                                 verify = 1;
1494                                 stat = 1;
1495                         } else if (!strcmp(arg, "--verify-stat-only")) {
1496                                 verify = 1;
1497                                 stat = 1;
1498                                 stat_only = 1;
1499                         } else if (!strcmp(arg, "--keep")) {
1500                                 keep_msg = "";
1501                         } else if (!prefixcmp(arg, "--keep=")) {
1502                                 keep_msg = arg + 7;
1503                         } else if (!prefixcmp(arg, "--threads=")) {
1504                                 char *end;
1505                                 nr_threads = strtoul(arg+10, &end, 0);
1506                                 if (!arg[10] || *end || nr_threads < 0)
1507                                         usage(index_pack_usage);
1508 #ifdef NO_PTHREADS
1509                                 if (nr_threads != 1)
1510                                         warning("no threads support, "
1511                                                 "ignoring %s", arg);
1512                                 nr_threads = 1;
1513 #endif
1514                         } else if (!prefixcmp(arg, "--pack_header=")) {
1515                                 struct pack_header *hdr;
1516                                 char *c;
1517
1518                                 hdr = (struct pack_header *)input_buffer;
1519                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1520                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1521                                 if (*c != ',')
1522                                         die(_("bad %s"), arg);
1523                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1524                                 if (*c)
1525                                         die(_("bad %s"), arg);
1526                                 input_len = sizeof(*hdr);
1527                         } else if (!strcmp(arg, "-v")) {
1528                                 verbose = 1;
1529                         } else if (!strcmp(arg, "-o")) {
1530                                 if (index_name || (i+1) >= argc)
1531                                         usage(index_pack_usage);
1532                                 index_name = argv[++i];
1533                         } else if (!prefixcmp(arg, "--index-version=")) {
1534                                 char *c;
1535                                 opts.version = strtoul(arg + 16, &c, 10);
1536                                 if (opts.version > 2)
1537                                         die(_("bad %s"), arg);
1538                                 if (*c == ',')
1539                                         opts.off32_limit = strtoul(c+1, &c, 0);
1540                                 if (*c || opts.off32_limit & 0x80000000)
1541                                         die(_("bad %s"), arg);
1542                         } else
1543                                 usage(index_pack_usage);
1544                         continue;
1545                 }
1546
1547                 if (pack_name)
1548                         usage(index_pack_usage);
1549                 pack_name = arg;
1550         }
1551
1552         if (!pack_name && !from_stdin)
1553                 usage(index_pack_usage);
1554         if (fix_thin_pack && !from_stdin)
1555                 die(_("--fix-thin cannot be used without --stdin"));
1556         if (!index_name && pack_name) {
1557                 int len = strlen(pack_name);
1558                 if (!has_extension(pack_name, ".pack"))
1559                         die(_("packfile name '%s' does not end with '.pack'"),
1560                             pack_name);
1561                 index_name_buf = xmalloc(len);
1562                 memcpy(index_name_buf, pack_name, len - 5);
1563                 strcpy(index_name_buf + len - 5, ".idx");
1564                 index_name = index_name_buf;
1565         }
1566         if (keep_msg && !keep_name && pack_name) {
1567                 int len = strlen(pack_name);
1568                 if (!has_extension(pack_name, ".pack"))
1569                         die(_("packfile name '%s' does not end with '.pack'"),
1570                             pack_name);
1571                 keep_name_buf = xmalloc(len);
1572                 memcpy(keep_name_buf, pack_name, len - 5);
1573                 strcpy(keep_name_buf + len - 5, ".keep");
1574                 keep_name = keep_name_buf;
1575         }
1576         if (verify) {
1577                 if (!index_name)
1578                         die(_("--verify with no packfile name given"));
1579                 read_idx_option(&opts, index_name);
1580                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1581         }
1582         if (strict)
1583                 opts.flags |= WRITE_IDX_STRICT;
1584
1585 #ifndef NO_PTHREADS
1586         if (!nr_threads) {
1587                 nr_threads = online_cpus();
1588                 /* An experiment showed that more threads does not mean faster */
1589                 if (nr_threads > 3)
1590                         nr_threads = 3;
1591         }
1592 #endif
1593
1594         curr_pack = open_pack_file(pack_name);
1595         parse_pack_header();
1596         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1597         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1598         parse_pack_objects(pack_sha1);
1599         resolve_deltas();
1600         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1601         free(deltas);
1602         if (strict)
1603                 check_objects();
1604
1605         if (stat)
1606                 show_pack_info(stat_only);
1607
1608         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1609         for (i = 0; i < nr_objects; i++)
1610                 idx_objects[i] = &objects[i].idx;
1611         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1612         free(idx_objects);
1613
1614         if (!verify)
1615                 final(pack_name, curr_pack,
1616                       index_name, curr_index,
1617                       keep_name, keep_msg,
1618                       pack_sha1);
1619         else
1620                 close(input_fd);
1621         free(objects);
1622         free(index_name_buf);
1623         free(keep_name_buf);
1624         if (pack_name == NULL)
1625                 free((void *) curr_pack);
1626         if (index_name == NULL)
1627                 free((void *) curr_index);
1628
1629         return 0;
1630 }