git.oblomov.eu Git - git/blob - 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 struct thread_local {
  44 #ifndef NO_PTHREADS
  45         pthread_t thread;
  46 #endif
  47         struct base_data *base_cache;
  48         size_t base_cache_used;
  49         int pack_fd;
  50 };
  51
  52 /*
  53  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  54  * to memcmp() only the first 20 bytes.
  55  */
  56 #define UNION_BASE_SZ   20
  57
  58 #define FLAG_LINK (1u<<20)
  59 #define FLAG_CHECKED (1u<<21)
  60
  61 struct delta_entry {
  62         union delta_base base;
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct delta_entry *deltas;
  68 static struct thread_local nothread_data;
  69 static int nr_objects;
  70 static int nr_deltas;
  71 static int nr_resolved_deltas;
  72 static int nr_threads;
  73
  74 static int from_stdin;
  75 static int strict;
  76 static int do_fsck_object;
  77 static int verbose;
  78 static int show_stat;
  79 static int check_self_contained_and_connected;
  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;
  91 static const char *curr_pack;
  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_mutex_t deepest_delta_mutex;
 112 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 113 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 114
 115 static pthread_mutex_t type_cas_mutex;
 116 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 117 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 118
 119 static pthread_key_t key;
 120
 121 static inline void lock_mutex(pthread_mutex_t *mutex)
 122 {
 123         if (threads_active)
 124                 pthread_mutex_lock(mutex);
 125 }
 126
 127 static inline void unlock_mutex(pthread_mutex_t *mutex)
 128 {
 129         if (threads_active)
 130                 pthread_mutex_unlock(mutex);
 131 }
 132
 133 /*
 134  * Mutex and conditional variable can't be statically-initialized on Windows.
 135  */
 136 static void init_thread(void)
 137 {
 138         int i;
 139         init_recursive_mutex(&read_mutex);
 140         pthread_mutex_init(&counter_mutex, NULL);
 141         pthread_mutex_init(&work_mutex, NULL);
 142         pthread_mutex_init(&type_cas_mutex, NULL);
 143         if (show_stat)
 144                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 145         pthread_key_create(&key, NULL);
 146         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 147         for (i = 0; i < nr_threads; i++) {
 148                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 149                 if (thread_data[i].pack_fd == -1)
 150                         die_errno(_("unable to open %s"), curr_pack);
 151         }
 152
 153         threads_active = 1;
 154 }
 155
 156 static void cleanup_thread(void)
 157 {
 158         int i;
 159         if (!threads_active)
 160                 return;
 161         threads_active = 0;
 162         pthread_mutex_destroy(&read_mutex);
 163         pthread_mutex_destroy(&counter_mutex);
 164         pthread_mutex_destroy(&work_mutex);
 165         pthread_mutex_destroy(&type_cas_mutex);
 166         if (show_stat)
 167                 pthread_mutex_destroy(&deepest_delta_mutex);
 168         for (i = 0; i < nr_threads; i++)
 169                 close(thread_data[i].pack_fd);
 170         pthread_key_delete(key);
 171         free(thread_data);
 172 }
 173
 174 #else
 175
 176 #define read_lock()
 177 #define read_unlock()
 178
 179 #define counter_lock()
 180 #define counter_unlock()
 181
 182 #define work_lock()
 183 #define work_unlock()
 184
 185 #define deepest_delta_lock()
 186 #define deepest_delta_unlock()
 187
 188 #endif
 189
 190
 191 static int mark_link(struct object *obj, int type, void *data)
 192 {
 193         if (!obj)
 194                 return -1;
 195
 196         if (type != OBJ_ANY && obj->type != type)
 197                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 198
 199         obj->flags |= FLAG_LINK;
 200         return 0;
 201 }
 202
 203 /* The content of each linked object must have been checked
 204    or it must be already present in the object database */
 205 static unsigned check_object(struct object *obj)
 206 {
 207         if (!obj)
 208                 return 0;
 209
 210         if (!(obj->flags & FLAG_LINK))
 211                 return 0;
 212
 213         if (!(obj->flags & FLAG_CHECKED)) {
 214                 unsigned long size;
 215                 int type = sha1_object_info(obj->sha1, &size);
 216                 if (type <= 0)
 217                         die(_("did not receive expected object %s"),
 218                               sha1_to_hex(obj->sha1));
 219                 if (type != obj->type)
 220                         die(_("object %s: expected type %s, found %s"),
 221                             sha1_to_hex(obj->sha1),
 222                             typename(obj->type), typename(type));
 223                 obj->flags |= FLAG_CHECKED;
 224                 return 1;
 225         }
 226
 227         return 0;
 228 }
 229
 230 static unsigned check_objects(void)
 231 {
 232         unsigned i, max, foreign_nr = 0;
 233
 234         max = get_max_object_index();
 235         for (i = 0; i < max; i++)
 236                 foreign_nr += check_object(get_indexed_object(i));
 237         return foreign_nr;
 238 }
 239
 240
 241 /* Discard current buffer used content. */
 242 static void flush(void)
 243 {
 244         if (input_offset) {
 245                 if (output_fd >= 0)
 246                         write_or_die(output_fd, input_buffer, input_offset);
 247                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 248                 memmove(input_buffer, input_buffer + input_offset, input_len);
 249                 input_offset = 0;
 250         }
 251 }
 252
 253 /*
 254  * Make sure at least "min" bytes are available in the buffer, and
 255  * return the pointer to the buffer.
 256  */
 257 static void *fill(int min)
 258 {
 259         if (min <= input_len)
 260                 return input_buffer + input_offset;
 261         if (min > sizeof(input_buffer))
 262                 die(Q_("cannot fill %d byte",
 263                        "cannot fill %d bytes",
 264                        min),
 265                     min);
 266         flush();
 267         do {
 268                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 269                                 sizeof(input_buffer) - input_len);
 270                 if (ret <= 0) {
 271                         if (!ret)
 272                                 die(_("early EOF"));
 273                         die_errno(_("read error on input"));
 274                 }
 275                 input_len += ret;
 276                 if (from_stdin)
 277                         display_throughput(progress, consumed_bytes + input_len);
 278         } while (input_len < min);
 279         return input_buffer;
 280 }
 281
 282 static void use(int bytes)
 283 {
 284         if (bytes > input_len)
 285                 die(_("used more bytes than were available"));
 286         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 287         input_len -= bytes;
 288         input_offset += bytes;
 289
 290         /* make sure off_t is sufficiently large not to wrap */
 291         if (signed_add_overflows(consumed_bytes, bytes))
 292                 die(_("pack too large for current definition of off_t"));
 293         consumed_bytes += bytes;
 294 }
 295
 296 static const char *open_pack_file(const char *pack_name)
 297 {
 298         if (from_stdin) {
 299                 input_fd = 0;
 300                 if (!pack_name) {
 301                         static char tmp_file[PATH_MAX];
 302                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 303                                                 "pack/tmp_pack_XXXXXX");
 304                         pack_name = xstrdup(tmp_file);
 305                 } else
 306                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 307                 if (output_fd < 0)
 308                         die_errno(_("unable to create '%s'"), pack_name);
 309                 nothread_data.pack_fd = output_fd;
 310         } else {
 311                 input_fd = open(pack_name, O_RDONLY);
 312                 if (input_fd < 0)
 313                         die_errno(_("cannot open packfile '%s'"), pack_name);
 314                 output_fd = -1;
 315                 nothread_data.pack_fd = input_fd;
 316         }
 317         git_SHA1_Init(&input_ctx);
 318         return pack_name;
 319 }
 320
 321 static void parse_pack_header(void)
 322 {
 323         struct pack_header *hdr = fill(sizeof(struct pack_header));
 324
 325         /* Header consistency check */
 326         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 327                 die(_("pack signature mismatch"));
 328         if (!pack_version_ok(hdr->hdr_version))
 329                 die(_("pack version %"PRIu32" unsupported"),
 330                         ntohl(hdr->hdr_version));
 331
 332         nr_objects = ntohl(hdr->hdr_entries);
 333         use(sizeof(struct pack_header));
 334 }
 335
 336 static NORETURN void bad_object(unsigned long offset, const char *format,
 337                        ...) __attribute__((format (printf, 2, 3)));
 338
 339 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 340 {
 341         va_list params;
 342         char buf[1024];
 343
 344         va_start(params, format);
 345         vsnprintf(buf, sizeof(buf), format, params);
 346         va_end(params);
 347         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 348 }
 349
 350 static inline struct thread_local *get_thread_data(void)
 351 {
 352 #ifndef NO_PTHREADS
 353         if (threads_active)
 354                 return pthread_getspecific(key);
 355         assert(!threads_active &&
 356                "This should only be reached when all threads are gone");
 357 #endif
 358         return &nothread_data;
 359 }
 360
 361 #ifndef NO_PTHREADS
 362 static void set_thread_data(struct thread_local *data)
 363 {
 364         if (threads_active)
 365                 pthread_setspecific(key, data);
 366 }
 367 #endif
 368
 369 static struct base_data *alloc_base_data(void)
 370 {
 371         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 372         base->ref_last = -1;
 373         base->ofs_last = -1;
 374         return base;
 375 }
 376
 377 static void free_base_data(struct base_data *c)
 378 {
 379         if (c->data) {
 380                 free(c->data);
 381                 c->data = NULL;
 382                 get_thread_data()->base_cache_used -= c->size;
 383         }
 384 }
 385
 386 static void prune_base_data(struct base_data *retain)
 387 {
 388         struct base_data *b;
 389         struct thread_local *data = get_thread_data();
 390         for (b = data->base_cache;
 391              data->base_cache_used > delta_base_cache_limit && b;
 392              b = b->child) {
 393                 if (b->data && b != retain)
 394                         free_base_data(b);
 395         }
 396 }
 397
 398 static void link_base_data(struct base_data *base, struct base_data *c)
 399 {
 400         if (base)
 401                 base->child = c;
 402         else
 403                 get_thread_data()->base_cache = c;
 404
 405         c->base = base;
 406         c->child = NULL;
 407         if (c->data)
 408                 get_thread_data()->base_cache_used += c->size;
 409         prune_base_data(c);
 410 }
 411
 412 static void unlink_base_data(struct base_data *c)
 413 {
 414         struct base_data *base = c->base;
 415         if (base)
 416                 base->child = NULL;
 417         else
 418                 get_thread_data()->base_cache = NULL;
 419         free_base_data(c);
 420 }
 421
 422 static int is_delta_type(enum object_type type)
 423 {
 424         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 425 }
 426
 427 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 428                                enum object_type type, unsigned char *sha1)
 429 {
 430         static char fixed_buf[8192];
 431         int status;
 432         git_zstream stream;
 433         void *buf;
 434         git_SHA_CTX c;
 435         char hdr[32];
 436         int hdrlen;
 437
 438         if (!is_delta_type(type)) {
 439                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 440                 git_SHA1_Init(&c);
 441                 git_SHA1_Update(&c, hdr, hdrlen);
 442         } else
 443                 sha1 = NULL;
 444         if (type == OBJ_BLOB && size > big_file_threshold)
 445                 buf = fixed_buf;
 446         else
 447                 buf = xmalloc(size);
 448
 449         memset(&stream, 0, sizeof(stream));
 450         git_inflate_init(&stream);
 451         stream.next_out = buf;
 452         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 453
 454         do {
 455                 unsigned char *last_out = stream.next_out;
 456                 stream.next_in = fill(1);
 457                 stream.avail_in = input_len;
 458                 status = git_inflate(&stream, 0);
 459                 use(input_len - stream.avail_in);
 460                 if (sha1)
 461                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 462                 if (buf == fixed_buf) {
 463                         stream.next_out = buf;
 464                         stream.avail_out = sizeof(fixed_buf);
 465                 }
 466         } while (status == Z_OK);
 467         if (stream.total_out != size || status != Z_STREAM_END)
 468                 bad_object(offset, _("inflate returned %d"), status);
 469         git_inflate_end(&stream);
 470         if (sha1)
 471                 git_SHA1_Final(sha1, &c);
 472         return buf == fixed_buf ? NULL : buf;
 473 }
 474
 475 static void *unpack_raw_entry(struct object_entry *obj,
 476                               union delta_base *delta_base,
 477                               unsigned char *sha1)
 478 {
 479         unsigned char *p;
 480         unsigned long size, c;
 481         off_t base_offset;
 482         unsigned shift;
 483         void *data;
 484
 485         obj->idx.offset = consumed_bytes;
 486         input_crc32 = crc32(0, NULL, 0);
 487
 488         p = fill(1);
 489         c = *p;
 490         use(1);
 491         obj->type = (c >> 4) & 7;
 492         size = (c & 15);
 493         shift = 4;
 494         while (c & 0x80) {
 495                 p = fill(1);
 496                 c = *p;
 497                 use(1);
 498                 size += (c & 0x7f) << shift;
 499                 shift += 7;
 500         }
 501         obj->size = size;
 502
 503         switch (obj->type) {
 504         case OBJ_REF_DELTA:
 505                 hashcpy(delta_base->sha1, fill(20));
 506                 use(20);
 507                 break;
 508         case OBJ_OFS_DELTA:
 509                 memset(delta_base, 0, sizeof(*delta_base));
 510                 p = fill(1);
 511                 c = *p;
 512                 use(1);
 513                 base_offset = c & 127;
 514                 while (c & 128) {
 515                         base_offset += 1;
 516                         if (!base_offset || MSB(base_offset, 7))
 517                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 518                         p = fill(1);
 519                         c = *p;
 520                         use(1);
 521                         base_offset = (base_offset << 7) + (c & 127);
 522                 }
 523                 delta_base->offset = obj->idx.offset - base_offset;
 524                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 525                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 526                 break;
 527         case OBJ_COMMIT:
 528         case OBJ_TREE:
 529         case OBJ_BLOB:
 530         case OBJ_TAG:
 531                 break;
 532         default:
 533                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 534         }
 535         obj->hdr_size = consumed_bytes - obj->idx.offset;
 536
 537         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 538         obj->idx.crc32 = input_crc32;
 539         return data;
 540 }
 541
 542 static void *unpack_data(struct object_entry *obj,
 543                          int (*consume)(const unsigned char *, unsigned long, void *),
 544                          void *cb_data)
 545 {
 546         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 547         unsigned long len = obj[1].idx.offset - from;
 548         unsigned char *data, *inbuf;
 549         git_zstream stream;
 550         int status;
 551
 552         data = xmalloc(consume ? 64*1024 : obj->size);
 553         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 554
 555         memset(&stream, 0, sizeof(stream));
 556         git_inflate_init(&stream);
 557         stream.next_out = data;
 558         stream.avail_out = consume ? 64*1024 : obj->size;
 559
 560         do {
 561                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 562                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 563                 if (n < 0)
 564                         die_errno(_("cannot pread pack file"));
 565                 if (!n)
 566                         die(Q_("premature end of pack file, %lu byte missing",
 567                                "premature end of pack file, %lu bytes missing",
 568                                len),
 569                             len);
 570                 from += n;
 571                 len -= n;
 572                 stream.next_in = inbuf;
 573                 stream.avail_in = n;
 574                 if (!consume)
 575                         status = git_inflate(&stream, 0);
 576                 else {
 577                         do {
 578                                 status = git_inflate(&stream, 0);
 579                                 if (consume(data, stream.next_out - data, cb_data)) {
 580                                         free(inbuf);
 581                                         free(data);
 582                                         return NULL;
 583                                 }
 584                                 stream.next_out = data;
 585                                 stream.avail_out = 64*1024;
 586                         } while (status == Z_OK && stream.avail_in);
 587                 }
 588         } while (len && status == Z_OK && !stream.avail_in);
 589
 590         /* This has been inflated OK when first encountered, so... */
 591         if (status != Z_STREAM_END || stream.total_out != obj->size)
 592                 die(_("serious inflate inconsistency"));
 593
 594         git_inflate_end(&stream);
 595         free(inbuf);
 596         if (consume) {
 597                 free(data);
 598                 data = NULL;
 599         }
 600         return data;
 601 }
 602
 603 static void *get_data_from_pack(struct object_entry *obj)
 604 {
 605         return unpack_data(obj, NULL, NULL);
 606 }
 607
 608 static int compare_delta_bases(const union delta_base *base1,
 609                                const union delta_base *base2,
 610                                enum object_type type1,
 611                                enum object_type type2)
 612 {
 613         int cmp = type1 - type2;
 614         if (cmp)
 615                 return cmp;
 616         return memcmp(base1, base2, UNION_BASE_SZ);
 617 }
 618
 619 static int find_delta(const union delta_base *base, enum object_type type)
 620 {
 621         int first = 0, last = nr_deltas;
 622
 623         while (first < last) {
 624                 int next = (first + last) / 2;
 625                 struct delta_entry *delta = &deltas[next];
 626                 int cmp;
 627
 628                 cmp = compare_delta_bases(base, &delta->base,
 629                                           type, objects[delta->obj_no].type);
 630                 if (!cmp)
 631                         return next;
 632                 if (cmp < 0) {
 633                         last = next;
 634                         continue;
 635                 }
 636                 first = next+1;
 637         }
 638         return -first-1;
 639 }
 640
 641 static void find_delta_children(const union delta_base *base,
 642                                 int *first_index, int *last_index,
 643                                 enum object_type type)
 644 {
 645         int first = find_delta(base, type);
 646         int last = first;
 647         int end = nr_deltas - 1;
 648
 649         if (first < 0) {
 650                 *first_index = 0;
 651                 *last_index = -1;
 652                 return;
 653         }
 654         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 655                 --first;
 656         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 657                 ++last;
 658         *first_index = first;
 659         *last_index = last;
 660 }
 661
 662 struct compare_data {
 663         struct object_entry *entry;
 664         struct git_istream *st;
 665         unsigned char *buf;
 666         unsigned long buf_size;
 667 };
 668
 669 static int compare_objects(const unsigned char *buf, unsigned long size,
 670                            void *cb_data)
 671 {
 672         struct compare_data *data = cb_data;
 673
 674         if (data->buf_size < size) {
 675                 free(data->buf);
 676                 data->buf = xmalloc(size);
 677                 data->buf_size = size;
 678         }
 679
 680         while (size) {
 681                 ssize_t len = read_istream(data->st, data->buf, size);
 682                 if (len == 0)
 683                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 684                             sha1_to_hex(data->entry->idx.sha1));
 685                 if (len < 0)
 686                         die(_("unable to read %s"),
 687                             sha1_to_hex(data->entry->idx.sha1));
 688                 if (memcmp(buf, data->buf, len))
 689                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 690                             sha1_to_hex(data->entry->idx.sha1));
 691                 size -= len;
 692                 buf += len;
 693         }
 694         return 0;
 695 }
 696
 697 static int check_collison(struct object_entry *entry)
 698 {
 699         struct compare_data data;
 700         enum object_type type;
 701         unsigned long size;
 702
 703         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 704                 return -1;
 705
 706         memset(&data, 0, sizeof(data));
 707         data.entry = entry;
 708         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 709         if (!data.st)
 710                 return -1;
 711         if (size != entry->size || type != entry->type)
 712                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 713                     sha1_to_hex(entry->idx.sha1));
 714         unpack_data(entry, compare_objects, &data);
 715         close_istream(data.st);
 716         free(data.buf);
 717         return 0;
 718 }
 719
 720 static void sha1_object(const void *data, struct object_entry *obj_entry,
 721                         unsigned long size, enum object_type type,
 722                         const unsigned char *sha1)
 723 {
 724         void *new_data = NULL;
 725         int collision_test_needed;
 726
 727         assert(data || obj_entry);
 728
 729         read_lock();
 730         collision_test_needed = has_sha1_file(sha1);
 731         read_unlock();
 732
 733         if (collision_test_needed && !data) {
 734                 read_lock();
 735                 if (!check_collison(obj_entry))
 736                         collision_test_needed = 0;
 737                 read_unlock();
 738         }
 739         if (collision_test_needed) {
 740                 void *has_data;
 741                 enum object_type has_type;
 742                 unsigned long has_size;
 743                 read_lock();
 744                 has_type = sha1_object_info(sha1, &has_size);
 745                 if (has_type != type || has_size != size)
 746                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 747                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 748                 read_unlock();
 749                 if (!data)
 750                         data = new_data = get_data_from_pack(obj_entry);
 751                 if (!has_data)
 752                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 753                 if (size != has_size || type != has_type ||
 754                     memcmp(data, has_data, size) != 0)
 755                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 756                 free(has_data);
 757         }
 758
 759         if (strict) {
 760                 read_lock();
 761                 if (type == OBJ_BLOB) {
 762                         struct blob *blob = lookup_blob(sha1);
 763                         if (blob)
 764                                 blob->object.flags |= FLAG_CHECKED;
 765                         else
 766                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 767                 } else {
 768                         struct object *obj;
 769                         int eaten;
 770                         void *buf = (void *) data;
 771
 772                         assert(data && "data can only be NULL for large _blobs_");
 773
 774                         /*
 775                          * we do not need to free the memory here, as the
 776                          * buf is deleted by the caller.
 777                          */
 778                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 779                         if (!obj)
 780                                 die(_("invalid %s"), typename(type));
 781                         if (do_fsck_object &&
 782                             fsck_object(obj, 1, fsck_error_function))
 783                                 die(_("Error in object"));
 784                         if (fsck_walk(obj, mark_link, NULL))
 785                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 786
 787                         if (obj->type == OBJ_TREE) {
 788                                 struct tree *item = (struct tree *) obj;
 789                                 item->buffer = NULL;
 790                                 obj->parsed = 0;
 791                         }
 792                         if (obj->type == OBJ_COMMIT) {
 793                                 struct commit *commit = (struct commit *) obj;
 794                                 if (detach_commit_buffer(commit, NULL) != data)
 795                                         die("BUG: parse_object_buffer transmogrified our buffer");
 796                         }
 797                         obj->flags |= FLAG_CHECKED;
 798                 }
 799                 read_unlock();
 800         }
 801
 802         free(new_data);
 803 }
 804
 805 /*
 806  * This function is part of find_unresolved_deltas(). There are two
 807  * walkers going in the opposite ways.
 808  *
 809  * The first one in find_unresolved_deltas() traverses down from
 810  * parent node to children, deflating nodes along the way. However,
 811  * memory for deflated nodes is limited by delta_base_cache_limit, so
 812  * at some point parent node's deflated content may be freed.
 813  *
 814  * The second walker is this function, which goes from current node up
 815  * to top parent if necessary to deflate the node. In normal
 816  * situation, its parent node would be already deflated, so it just
 817  * needs to apply delta.
 818  *
 819  * In the worst case scenario, parent node is no longer deflated because
 820  * we're running out of delta_base_cache_limit; we need to re-deflate
 821  * parents, possibly up to the top base.
 822  *
 823  * All deflated objects here are subject to be freed if we exceed
 824  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 825  * just need to make sure the last node is not freed.
 826  */
 827 static void *get_base_data(struct base_data *c)
 828 {
 829         if (!c->data) {
 830                 struct object_entry *obj = c->obj;
 831                 struct base_data **delta = NULL;
 832                 int delta_nr = 0, delta_alloc = 0;
 833
 834                 while (is_delta_type(c->obj->type) && !c->data) {
 835                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 836                         delta[delta_nr++] = c;
 837                         c = c->base;
 838                 }
 839                 if (!delta_nr) {
 840                         c->data = get_data_from_pack(obj);
 841                         c->size = obj->size;
 842                         get_thread_data()->base_cache_used += c->size;
 843                         prune_base_data(c);
 844                 }
 845                 for (; delta_nr > 0; delta_nr--) {
 846                         void *base, *raw;
 847                         c = delta[delta_nr - 1];
 848                         obj = c->obj;
 849                         base = get_base_data(c->base);
 850                         raw = get_data_from_pack(obj);
 851                         c->data = patch_delta(
 852                                 base, c->base->size,
 853                                 raw, obj->size,
 854                                 &c->size);
 855                         free(raw);
 856                         if (!c->data)
 857                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 858                         get_thread_data()->base_cache_used += c->size;
 859                         prune_base_data(c);
 860                 }
 861                 free(delta);
 862         }
 863         return c->data;
 864 }
 865
 866 static void resolve_delta(struct object_entry *delta_obj,
 867                           struct base_data *base, struct base_data *result)
 868 {
 869         void *base_data, *delta_data;
 870
 871         if (show_stat) {
 872                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 873                 deepest_delta_lock();
 874                 if (deepest_delta < delta_obj->delta_depth)
 875                         deepest_delta = delta_obj->delta_depth;
 876                 deepest_delta_unlock();
 877         }
 878         delta_obj->base_object_no = base->obj - objects;
 879         delta_data = get_data_from_pack(delta_obj);
 880         base_data = get_base_data(base);
 881         result->obj = delta_obj;
 882         result->data = patch_delta(base_data, base->size,
 883                                    delta_data, delta_obj->size, &result->size);
 884         free(delta_data);
 885         if (!result->data)
 886                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 887         hash_sha1_file(result->data, result->size,
 888                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 889         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 890                     delta_obj->idx.sha1);
 891         counter_lock();
 892         nr_resolved_deltas++;
 893         counter_unlock();
 894 }
 895
 896 /*
 897  * Standard boolean compare-and-swap: atomically check whether "*type" is
 898  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 899  * and return false.
 900  */
 901 static int compare_and_swap_type(enum object_type *type,
 902                                  enum object_type want,
 903                                  enum object_type set)
 904 {
 905         enum object_type old;
 906
 907         type_cas_lock();
 908         old = *type;
 909         if (old == want)
 910                 *type = set;
 911         type_cas_unlock();
 912
 913         return old == want;
 914 }
 915
 916 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 917                                                   struct base_data *prev_base)
 918 {
 919         if (base->ref_last == -1 && base->ofs_last == -1) {
 920                 union delta_base base_spec;
 921
 922                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 923                 find_delta_children(&base_spec,
 924                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 925
 926                 memset(&base_spec, 0, sizeof(base_spec));
 927                 base_spec.offset = base->obj->idx.offset;
 928                 find_delta_children(&base_spec,
 929                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 930
 931                 if (base->ref_last == -1 && base->ofs_last == -1) {
 932                         free(base->data);
 933                         return NULL;
 934                 }
 935
 936                 link_base_data(prev_base, base);
 937         }
 938
 939         if (base->ref_first <= base->ref_last) {
 940                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 941                 struct base_data *result = alloc_base_data();
 942
 943                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
 944                                            base->obj->real_type))
 945                         die("BUG: child->real_type != OBJ_REF_DELTA");
 946
 947                 resolve_delta(child, base, result);
 948                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 949                         free_base_data(base);
 950
 951                 base->ref_first++;
 952                 return result;
 953         }
 954
 955         if (base->ofs_first <= base->ofs_last) {
 956                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 957                 struct base_data *result = alloc_base_data();
 958
 959                 assert(child->real_type == OBJ_OFS_DELTA);
 960                 child->real_type = base->obj->real_type;
 961                 resolve_delta(child, base, result);
 962                 if (base->ofs_first == base->ofs_last)
 963                         free_base_data(base);
 964
 965                 base->ofs_first++;
 966                 return result;
 967         }
 968
 969         unlink_base_data(base);
 970         return NULL;
 971 }
 972
 973 static void find_unresolved_deltas(struct base_data *base)
 974 {
 975         struct base_data *new_base, *prev_base = NULL;
 976         for (;;) {
 977                 new_base = find_unresolved_deltas_1(base, prev_base);
 978
 979                 if (new_base) {
 980                         prev_base = base;
 981                         base = new_base;
 982                 } else {
 983                         free(base);
 984                         base = prev_base;
 985                         if (!base)
 986                                 return;
 987                         prev_base = base->base;
 988                 }
 989         }
 990 }
 991
 992 static int compare_delta_entry(const void *a, const void *b)
 993 {
 994         const struct delta_entry *delta_a = a;
 995         const struct delta_entry *delta_b = b;
 996
 997         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 998         return compare_delta_bases(&delta_a->base, &delta_b->base,
 999                                    objects[delta_a->obj_no].type,
1000                                    objects[delta_b->obj_no].type);
1001 }
1002
1003 static void resolve_base(struct object_entry *obj)
1004 {
1005         struct base_data *base_obj = alloc_base_data();
1006         base_obj->obj = obj;
1007         base_obj->data = NULL;
1008         find_unresolved_deltas(base_obj);
1009 }
1010
1011 #ifndef NO_PTHREADS
1012 static void *threaded_second_pass(void *data)
1013 {
1014         set_thread_data(data);
1015         for (;;) {
1016                 int i;
1017                 counter_lock();
1018                 display_progress(progress, nr_resolved_deltas);
1019                 counter_unlock();
1020                 work_lock();
1021                 while (nr_dispatched < nr_objects &&
1022                        is_delta_type(objects[nr_dispatched].type))
1023                         nr_dispatched++;
1024                 if (nr_dispatched >= nr_objects) {
1025                         work_unlock();
1026                         break;
1027                 }
1028                 i = nr_dispatched++;
1029                 work_unlock();
1030
1031                 resolve_base(&objects[i]);
1032         }
1033         return NULL;
1034 }
1035 #endif
1036
1037 /*
1038  * First pass:
1039  * - find locations of all objects;
1040  * - calculate SHA1 of all non-delta objects;
1041  * - remember base (SHA1 or offset) for all deltas.
1042  */
1043 static void parse_pack_objects(unsigned char *sha1)
1044 {
1045         int i, nr_delays = 0;
1046         struct delta_entry *delta = deltas;
1047         struct stat st;
1048
1049         if (verbose)
1050                 progress = start_progress(
1051                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1052                                 nr_objects);
1053         for (i = 0; i < nr_objects; i++) {
1054                 struct object_entry *obj = &objects[i];
1055                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1056                 obj->real_type = obj->type;
1057                 if (is_delta_type(obj->type)) {
1058                         nr_deltas++;
1059                         delta->obj_no = i;
1060                         delta++;
1061                 } else if (!data) {
1062                         /* large blobs, check later */
1063                         obj->real_type = OBJ_BAD;
1064                         nr_delays++;
1065                 } else
1066                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1067                 free(data);
1068                 display_progress(progress, i+1);
1069         }
1070         objects[i].idx.offset = consumed_bytes;
1071         stop_progress(&progress);
1072
1073         /* Check pack integrity */
1074         flush();
1075         git_SHA1_Final(sha1, &input_ctx);
1076         if (hashcmp(fill(20), sha1))
1077                 die(_("pack is corrupted (SHA1 mismatch)"));
1078         use(20);
1079
1080         /* If input_fd is a file, we should have reached its end now. */
1081         if (fstat(input_fd, &st))
1082                 die_errno(_("cannot fstat packfile"));
1083         if (S_ISREG(st.st_mode) &&
1084                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1085                 die(_("pack has junk at the end"));
1086
1087         for (i = 0; i < nr_objects; i++) {
1088                 struct object_entry *obj = &objects[i];
1089                 if (obj->real_type != OBJ_BAD)
1090                         continue;
1091                 obj->real_type = obj->type;
1092                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1093                 nr_delays--;
1094         }
1095         if (nr_delays)
1096                 die(_("confusion beyond insanity in parse_pack_objects()"));
1097 }
1098
1099 /*
1100  * Second pass:
1101  * - for all non-delta objects, look if it is used as a base for
1102  *   deltas;
1103  * - if used as a base, uncompress the object and apply all deltas,
1104  *   recursively checking if the resulting object is used as a base
1105  *   for some more deltas.
1106  */
1107 static void resolve_deltas(void)
1108 {
1109         int i;
1110
1111         if (!nr_deltas)
1112                 return;
1113
1114         /* Sort deltas by base SHA1/offset for fast searching */
1115         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1116               compare_delta_entry);
1117
1118         if (verbose)
1119                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1120
1121 #ifndef NO_PTHREADS
1122         nr_dispatched = 0;
1123         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1124                 init_thread();
1125                 for (i = 0; i < nr_threads; i++) {
1126                         int ret = pthread_create(&thread_data[i].thread, NULL,
1127                                                  threaded_second_pass, thread_data + i);
1128                         if (ret)
1129                                 die(_("unable to create thread: %s"),
1130                                     strerror(ret));
1131                 }
1132                 for (i = 0; i < nr_threads; i++)
1133                         pthread_join(thread_data[i].thread, NULL);
1134                 cleanup_thread();
1135                 return;
1136         }
1137 #endif
1138
1139         for (i = 0; i < nr_objects; i++) {
1140                 struct object_entry *obj = &objects[i];
1141
1142                 if (is_delta_type(obj->type))
1143                         continue;
1144                 resolve_base(obj);
1145                 display_progress(progress, nr_resolved_deltas);
1146         }
1147 }
1148
1149 /*
1150  * Third pass:
1151  * - append objects to convert thin pack to full pack if required
1152  * - write the final 20-byte SHA-1
1153  */
1154 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1155 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1156 {
1157         if (nr_deltas == nr_resolved_deltas) {
1158                 stop_progress(&progress);
1159                 /* Flush remaining pack final 20-byte SHA1. */
1160                 flush();
1161                 return;
1162         }
1163
1164         if (fix_thin_pack) {
1165                 struct sha1file *f;
1166                 unsigned char read_sha1[20], tail_sha1[20];
1167                 struct strbuf msg = STRBUF_INIT;
1168                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1169                 int nr_objects_initial = nr_objects;
1170                 if (nr_unresolved <= 0)
1171                         die(_("confusion beyond insanity"));
1172                 objects = xrealloc(objects,
1173                                    (nr_objects + nr_unresolved + 1)
1174                                    * sizeof(*objects));
1175                 memset(objects + nr_objects + 1, 0,
1176                        nr_unresolved * sizeof(*objects));
1177                 f = sha1fd(output_fd, curr_pack);
1178                 fix_unresolved_deltas(f, nr_unresolved);
1179                 strbuf_addf(&msg, _("completed with %d local objects"),
1180                             nr_objects - nr_objects_initial);
1181                 stop_progress_msg(&progress, msg.buf);
1182                 strbuf_release(&msg);
1183                 sha1close(f, tail_sha1, 0);
1184                 hashcpy(read_sha1, pack_sha1);
1185                 fixup_pack_header_footer(output_fd, pack_sha1,
1186                                          curr_pack, nr_objects,
1187                                          read_sha1, consumed_bytes-20);
1188                 if (hashcmp(read_sha1, tail_sha1) != 0)
1189                         die(_("Unexpected tail checksum for %s "
1190                               "(disk corruption?)"), curr_pack);
1191         }
1192         if (nr_deltas != nr_resolved_deltas)
1193                 die(Q_("pack has %d unresolved delta",
1194                        "pack has %d unresolved deltas",
1195                        nr_deltas - nr_resolved_deltas),
1196                     nr_deltas - nr_resolved_deltas);
1197 }
1198
1199 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1200 {
1201         git_zstream stream;
1202         int status;
1203         unsigned char outbuf[4096];
1204
1205         memset(&stream, 0, sizeof(stream));
1206         git_deflate_init(&stream, zlib_compression_level);
1207         stream.next_in = in;
1208         stream.avail_in = size;
1209
1210         do {
1211                 stream.next_out = outbuf;
1212                 stream.avail_out = sizeof(outbuf);
1213                 status = git_deflate(&stream, Z_FINISH);
1214                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1215         } while (status == Z_OK);
1216
1217         if (status != Z_STREAM_END)
1218                 die(_("unable to deflate appended object (%d)"), status);
1219         size = stream.total_out;
1220         git_deflate_end(&stream);
1221         return size;
1222 }
1223
1224 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1225                                const unsigned char *sha1, void *buf,
1226                                unsigned long size, enum object_type type)
1227 {
1228         struct object_entry *obj = &objects[nr_objects++];
1229         unsigned char header[10];
1230         unsigned long s = size;
1231         int n = 0;
1232         unsigned char c = (type << 4) | (s & 15);
1233         s >>= 4;
1234         while (s) {
1235                 header[n++] = c | 0x80;
1236                 c = s & 0x7f;
1237                 s >>= 7;
1238         }
1239         header[n++] = c;
1240         crc32_begin(f);
1241         sha1write(f, header, n);
1242         obj[0].size = size;
1243         obj[0].hdr_size = n;
1244         obj[0].type = type;
1245         obj[0].real_type = type;
1246         obj[1].idx.offset = obj[0].idx.offset + n;
1247         obj[1].idx.offset += write_compressed(f, buf, size);
1248         obj[0].idx.crc32 = crc32_end(f);
1249         sha1flush(f);
1250         hashcpy(obj->idx.sha1, sha1);
1251         return obj;
1252 }
1253
1254 static int delta_pos_compare(const void *_a, const void *_b)
1255 {
1256         struct delta_entry *a = *(struct delta_entry **)_a;
1257         struct delta_entry *b = *(struct delta_entry **)_b;
1258         return a->obj_no - b->obj_no;
1259 }
1260
1261 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1262 {
1263         struct delta_entry **sorted_by_pos;
1264         int i, n = 0;
1265
1266         /*
1267          * Since many unresolved deltas may well be themselves base objects
1268          * for more unresolved deltas, we really want to include the
1269          * smallest number of base objects that would cover as much delta
1270          * as possible by picking the
1271          * trunc deltas first, allowing for other deltas to resolve without
1272          * additional base objects.  Since most base objects are to be found
1273          * before deltas depending on them, a good heuristic is to start
1274          * resolving deltas in the same order as their position in the pack.
1275          */
1276         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1277         for (i = 0; i < nr_deltas; i++) {
1278                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1279                         continue;
1280                 sorted_by_pos[n++] = &deltas[i];
1281         }
1282         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1283
1284         for (i = 0; i < n; i++) {
1285                 struct delta_entry *d = sorted_by_pos[i];
1286                 enum object_type type;
1287                 struct base_data *base_obj = alloc_base_data();
1288
1289                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1290                         continue;
1291                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1292                 if (!base_obj->data)
1293                         continue;
1294
1295                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1296                                 base_obj->size, typename(type)))
1297                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1298                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1299                                         base_obj->data, base_obj->size, type);
1300                 find_unresolved_deltas(base_obj);
1301                 display_progress(progress, nr_resolved_deltas);
1302         }
1303         free(sorted_by_pos);
1304 }
1305
1306 static void final(const char *final_pack_name, const char *curr_pack_name,
1307                   const char *final_index_name, const char *curr_index_name,
1308                   const char *keep_name, const char *keep_msg,
1309                   unsigned char *sha1)
1310 {
1311         const char *report = "pack";
1312         char name[PATH_MAX];
1313         int err;
1314
1315         if (!from_stdin) {
1316                 close(input_fd);
1317         } else {
1318                 fsync_or_die(output_fd, curr_pack_name);
1319                 err = close(output_fd);
1320                 if (err)
1321                         die_errno(_("error while closing pack file"));
1322         }
1323
1324         if (keep_msg) {
1325                 int keep_fd, keep_msg_len = strlen(keep_msg);
1326
1327                 if (!keep_name)
1328                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1329                 else
1330                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1331
1332                 if (keep_fd < 0) {
1333                         if (errno != EEXIST)
1334                                 die_errno(_("cannot write keep file '%s'"),
1335                                           keep_name ? keep_name : name);
1336                 } else {
1337                         if (keep_msg_len > 0) {
1338                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1339                                 write_or_die(keep_fd, "\n", 1);
1340                         }
1341                         if (close(keep_fd) != 0)
1342                                 die_errno(_("cannot close written keep file '%s'"),
1343                                           keep_name ? keep_name : name);
1344                         report = "keep";
1345                 }
1346         }
1347
1348         if (final_pack_name != curr_pack_name) {
1349                 if (!final_pack_name) {
1350                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1351                                  get_object_directory(), sha1_to_hex(sha1));
1352                         final_pack_name = name;
1353                 }
1354                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1355                         die(_("cannot store pack file"));
1356         } else if (from_stdin)
1357                 chmod(final_pack_name, 0444);
1358
1359         if (final_index_name != curr_index_name) {
1360                 if (!final_index_name) {
1361                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1362                                  get_object_directory(), sha1_to_hex(sha1));
1363                         final_index_name = name;
1364                 }
1365                 if (move_temp_to_file(curr_index_name, final_index_name))
1366                         die(_("cannot store index file"));
1367         } else
1368                 chmod(final_index_name, 0444);
1369
1370         if (!from_stdin) {
1371                 printf("%s\n", sha1_to_hex(sha1));
1372         } else {
1373                 char buf[48];
1374                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1375                                    report, sha1_to_hex(sha1));
1376                 write_or_die(1, buf, len);
1377
1378                 /*
1379                  * Let's just mimic git-unpack-objects here and write
1380                  * the last part of the input buffer to stdout.
1381                  */
1382                 while (input_len) {
1383                         err = xwrite(1, input_buffer + input_offset, input_len);
1384                         if (err <= 0)
1385                                 break;
1386                         input_len -= err;
1387                         input_offset += err;
1388                 }
1389         }
1390 }
1391
1392 static int git_index_pack_config(const char *k, const char *v, void *cb)
1393 {
1394         struct pack_idx_option *opts = cb;
1395
1396         if (!strcmp(k, "pack.indexversion")) {
1397                 opts->version = git_config_int(k, v);
1398                 if (opts->version > 2)
1399                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1400                 return 0;
1401         }
1402         if (!strcmp(k, "pack.threads")) {
1403                 nr_threads = git_config_int(k, v);
1404                 if (nr_threads < 0)
1405                         die(_("invalid number of threads specified (%d)"),
1406                             nr_threads);
1407 #ifdef NO_PTHREADS
1408                 if (nr_threads != 1)
1409                         warning(_("no threads support, ignoring %s"), k);
1410                 nr_threads = 1;
1411 #endif
1412                 return 0;
1413         }
1414         return git_default_config(k, v, cb);
1415 }
1416
1417 static int cmp_uint32(const void *a_, const void *b_)
1418 {
1419         uint32_t a = *((uint32_t *)a_);
1420         uint32_t b = *((uint32_t *)b_);
1421
1422         return (a < b) ? -1 : (a != b);
1423 }
1424
1425 static void read_v2_anomalous_offsets(struct packed_git *p,
1426                                       struct pack_idx_option *opts)
1427 {
1428         const uint32_t *idx1, *idx2;
1429         uint32_t i;
1430
1431         /* The address of the 4-byte offset table */
1432         idx1 = (((const uint32_t *)p->index_data)
1433                 + 2 /* 8-byte header */
1434                 + 256 /* fan out */
1435                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1436                 + p->num_objects /* CRC32 table */
1437                 );
1438
1439         /* The address of the 8-byte offset table */
1440         idx2 = idx1 + p->num_objects;
1441
1442         for (i = 0; i < p->num_objects; i++) {
1443                 uint32_t off = ntohl(idx1[i]);
1444                 if (!(off & 0x80000000))
1445                         continue;
1446                 off = off & 0x7fffffff;
1447                 if (idx2[off * 2])
1448                         continue;
1449                 /*
1450                  * The real offset is ntohl(idx2[off * 2]) in high 4
1451                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1452                  * octets.  But idx2[off * 2] is Zero!!!
1453                  */
1454                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1455                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1456         }
1457
1458         if (1 < opts->anomaly_nr)
1459                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1460 }
1461
1462 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1463 {
1464         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1465
1466         if (!p)
1467                 die(_("Cannot open existing pack file '%s'"), pack_name);
1468         if (open_pack_index(p))
1469                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1470
1471         /* Read the attributes from the existing idx file */
1472         opts->version = p->index_version;
1473
1474         if (opts->version == 2)
1475                 read_v2_anomalous_offsets(p, opts);
1476
1477         /*
1478          * Get rid of the idx file as we do not need it anymore.
1479          * NEEDSWORK: extract this bit from free_pack_by_name() in
1480          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1481          * know we haven't installed this pack (hence we never have
1482          * read anything from it).
1483          */
1484         close_pack_index(p);
1485         free(p);
1486 }
1487
1488 static void show_pack_info(int stat_only)
1489 {
1490         int i, baseobjects = nr_objects - nr_deltas;
1491         unsigned long *chain_histogram = NULL;
1492
1493         if (deepest_delta)
1494                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1495
1496         for (i = 0; i < nr_objects; i++) {
1497                 struct object_entry *obj = &objects[i];
1498
1499                 if (is_delta_type(obj->type))
1500                         chain_histogram[obj->delta_depth - 1]++;
1501                 if (stat_only)
1502                         continue;
1503                 printf("%s %-6s %lu %lu %"PRIuMAX,
1504                        sha1_to_hex(obj->idx.sha1),
1505                        typename(obj->real_type), obj->size,
1506                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1507                        (uintmax_t)obj->idx.offset);
1508                 if (is_delta_type(obj->type)) {
1509                         struct object_entry *bobj = &objects[obj->base_object_no];
1510                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1511                 }
1512                 putchar('\n');
1513         }
1514
1515         if (baseobjects)
1516                 printf_ln(Q_("non delta: %d object",
1517                              "non delta: %d objects",
1518                              baseobjects),
1519                           baseobjects);
1520         for (i = 0; i < deepest_delta; i++) {
1521                 if (!chain_histogram[i])
1522                         continue;
1523                 printf_ln(Q_("chain length = %d: %lu object",
1524                              "chain length = %d: %lu objects",
1525                              chain_histogram[i]),
1526                           i + 1,
1527                           chain_histogram[i]);
1528         }
1529 }
1530
1531 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1532 {
1533         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1534         const char *curr_index;
1535         const char *index_name = NULL, *pack_name = NULL;
1536         const char *keep_name = NULL, *keep_msg = NULL;
1537         struct strbuf index_name_buf = STRBUF_INIT,
1538                       keep_name_buf = STRBUF_INIT;
1539         struct pack_idx_entry **idx_objects;
1540         struct pack_idx_option opts;
1541         unsigned char pack_sha1[20];
1542         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1543
1544         if (argc == 2 && !strcmp(argv[1], "-h"))
1545                 usage(index_pack_usage);
1546
1547         check_replace_refs = 0;
1548
1549         reset_pack_idx_option(&opts);
1550         git_config(git_index_pack_config, &opts);
1551         if (prefix && chdir(prefix))
1552                 die(_("Cannot come back to cwd"));
1553
1554         for (i = 1; i < argc; i++) {
1555                 const char *arg = argv[i];
1556
1557                 if (*arg == '-') {
1558                         if (!strcmp(arg, "--stdin")) {
1559                                 from_stdin = 1;
1560                         } else if (!strcmp(arg, "--fix-thin")) {
1561                                 fix_thin_pack = 1;
1562                         } else if (!strcmp(arg, "--strict")) {
1563                                 strict = 1;
1564                                 do_fsck_object = 1;
1565                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1566                                 strict = 1;
1567                                 check_self_contained_and_connected = 1;
1568                         } else if (!strcmp(arg, "--verify")) {
1569                                 verify = 1;
1570                         } else if (!strcmp(arg, "--verify-stat")) {
1571                                 verify = 1;
1572                                 show_stat = 1;
1573                         } else if (!strcmp(arg, "--verify-stat-only")) {
1574                                 verify = 1;
1575                                 show_stat = 1;
1576                                 stat_only = 1;
1577                         } else if (!strcmp(arg, "--keep")) {
1578                                 keep_msg = "";
1579                         } else if (starts_with(arg, "--keep=")) {
1580                                 keep_msg = arg + 7;
1581                         } else if (starts_with(arg, "--threads=")) {
1582                                 char *end;
1583                                 nr_threads = strtoul(arg+10, &end, 0);
1584                                 if (!arg[10] || *end || nr_threads < 0)
1585                                         usage(index_pack_usage);
1586 #ifdef NO_PTHREADS
1587                                 if (nr_threads != 1)
1588                                         warning(_("no threads support, "
1589                                                   "ignoring %s"), arg);
1590                                 nr_threads = 1;
1591 #endif
1592                         } else if (starts_with(arg, "--pack_header=")) {
1593                                 struct pack_header *hdr;
1594                                 char *c;
1595
1596                                 hdr = (struct pack_header *)input_buffer;
1597                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1598                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1599                                 if (*c != ',')
1600                                         die(_("bad %s"), arg);
1601                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1602                                 if (*c)
1603                                         die(_("bad %s"), arg);
1604                                 input_len = sizeof(*hdr);
1605                         } else if (!strcmp(arg, "-v")) {
1606                                 verbose = 1;
1607                         } else if (!strcmp(arg, "-o")) {
1608                                 if (index_name || (i+1) >= argc)
1609                                         usage(index_pack_usage);
1610                                 index_name = argv[++i];
1611                         } else if (starts_with(arg, "--index-version=")) {
1612                                 char *c;
1613                                 opts.version = strtoul(arg + 16, &c, 10);
1614                                 if (opts.version > 2)
1615                                         die(_("bad %s"), arg);
1616                                 if (*c == ',')
1617                                         opts.off32_limit = strtoul(c+1, &c, 0);
1618                                 if (*c || opts.off32_limit & 0x80000000)
1619                                         die(_("bad %s"), arg);
1620                         } else
1621                                 usage(index_pack_usage);
1622                         continue;
1623                 }
1624
1625                 if (pack_name)
1626                         usage(index_pack_usage);
1627                 pack_name = arg;
1628         }
1629
1630         if (!pack_name && !from_stdin)
1631                 usage(index_pack_usage);
1632         if (fix_thin_pack && !from_stdin)
1633                 die(_("--fix-thin cannot be used without --stdin"));
1634         if (!index_name && pack_name) {
1635                 size_t len;
1636                 if (!strip_suffix(pack_name, ".pack", &len))
1637                         die(_("packfile name '%s' does not end with '.pack'"),
1638                             pack_name);
1639                 strbuf_add(&index_name_buf, pack_name, len);
1640                 strbuf_addstr(&index_name_buf, ".idx");
1641                 index_name = index_name_buf.buf;
1642         }
1643         if (keep_msg && !keep_name && pack_name) {
1644                 size_t len;
1645                 if (!strip_suffix(pack_name, ".pack", &len))
1646                         die(_("packfile name '%s' does not end with '.pack'"),
1647                             pack_name);
1648                 strbuf_add(&keep_name_buf, pack_name, len);
1649                 strbuf_addstr(&keep_name_buf, ".idx");
1650                 keep_name = keep_name_buf.buf;
1651         }
1652         if (verify) {
1653                 if (!index_name)
1654                         die(_("--verify with no packfile name given"));
1655                 read_idx_option(&opts, index_name);
1656                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1657         }
1658         if (strict)
1659                 opts.flags |= WRITE_IDX_STRICT;
1660
1661 #ifndef NO_PTHREADS
1662         if (!nr_threads) {
1663                 nr_threads = online_cpus();
1664                 /* An experiment showed that more threads does not mean faster */
1665                 if (nr_threads > 3)
1666                         nr_threads = 3;
1667         }
1668 #endif
1669
1670         curr_pack = open_pack_file(pack_name);
1671         parse_pack_header();
1672         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1673         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1674         parse_pack_objects(pack_sha1);
1675         resolve_deltas();
1676         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1677         free(deltas);
1678         if (strict)
1679                 foreign_nr = check_objects();
1680
1681         if (show_stat)
1682                 show_pack_info(stat_only);
1683
1684         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1685         for (i = 0; i < nr_objects; i++)
1686                 idx_objects[i] = &objects[i].idx;
1687         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1688         free(idx_objects);
1689
1690         if (!verify)
1691                 final(pack_name, curr_pack,
1692                       index_name, curr_index,
1693                       keep_name, keep_msg,
1694                       pack_sha1);
1695         else
1696                 close(input_fd);
1697         free(objects);
1698         strbuf_release(&index_name_buf);
1699         strbuf_release(&keep_name_buf);
1700         if (pack_name == NULL)
1701                 free((void *) curr_pack);
1702         if (index_name == NULL)
1703                 free((void *) curr_index);
1704
1705         /*
1706          * Let the caller know this pack is not self contained
1707          */
1708         if (check_self_contained_and_connected && foreign_nr)
1709                 return 1;
1710
1711         return 0;
1712 }