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 #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 static int show_stat;
  82
  83 static struct progress *progress;
  84
  85 /* We always read in 4kB chunks. */
  86 static unsigned char input_buffer[4096];
  87 static unsigned int input_offset, input_len;
  88 static off_t consumed_bytes;
  89 static unsigned deepest_delta;
  90 static git_SHA_CTX input_ctx;
  91 static uint32_t input_crc32;
  92 static int input_fd, output_fd, pack_fd;
  93
  94 #ifndef NO_PTHREADS
  95
  96 static struct thread_local *thread_data;
  97 static int nr_dispatched;
  98 static int threads_active;
  99
 100 static pthread_mutex_t read_mutex;
 101 #define read_lock()             lock_mutex(&read_mutex)
 102 #define read_unlock()           unlock_mutex(&read_mutex)
 103
 104 static pthread_mutex_t counter_mutex;
 105 #define counter_lock()          lock_mutex(&counter_mutex)
 106 #define counter_unlock()        unlock_mutex(&counter_mutex)
 107
 108 static pthread_mutex_t work_mutex;
 109 #define work_lock()             lock_mutex(&work_mutex)
 110 #define work_unlock()           unlock_mutex(&work_mutex)
 111
 112 static pthread_mutex_t deepest_delta_mutex;
 113 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 114 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 115
 116 static pthread_key_t key;
 117
 118 static inline void lock_mutex(pthread_mutex_t *mutex)
 119 {
 120         if (threads_active)
 121                 pthread_mutex_lock(mutex);
 122 }
 123
 124 static inline void unlock_mutex(pthread_mutex_t *mutex)
 125 {
 126         if (threads_active)
 127                 pthread_mutex_unlock(mutex);
 128 }
 129
 130 /*
 131  * Mutex and conditional variable can't be statically-initialized on Windows.
 132  */
 133 static void init_thread(void)
 134 {
 135         init_recursive_mutex(&read_mutex);
 136         pthread_mutex_init(&counter_mutex, NULL);
 137         pthread_mutex_init(&work_mutex, NULL);
 138         if (show_stat)
 139                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 140         pthread_key_create(&key, NULL);
 141         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 142         threads_active = 1;
 143 }
 144
 145 static void cleanup_thread(void)
 146 {
 147         if (!threads_active)
 148                 return;
 149         threads_active = 0;
 150         pthread_mutex_destroy(&read_mutex);
 151         pthread_mutex_destroy(&counter_mutex);
 152         pthread_mutex_destroy(&work_mutex);
 153         if (show_stat)
 154                 pthread_mutex_destroy(&deepest_delta_mutex);
 155         pthread_key_delete(key);
 156         free(thread_data);
 157 }
 158
 159 #else
 160
 161 #define read_lock()
 162 #define read_unlock()
 163
 164 #define counter_lock()
 165 #define counter_unlock()
 166
 167 #define work_lock()
 168 #define work_unlock()
 169
 170 #define deepest_delta_lock()
 171 #define deepest_delta_unlock()
 172
 173 #endif
 174
 175
 176 static int mark_link(struct object *obj, int type, void *data)
 177 {
 178         if (!obj)
 179                 return -1;
 180
 181         if (type != OBJ_ANY && obj->type != type)
 182                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 183
 184         obj->flags |= FLAG_LINK;
 185         return 0;
 186 }
 187
 188 /* The content of each linked object must have been checked
 189    or it must be already present in the object database */
 190 static void check_object(struct object *obj)
 191 {
 192         if (!obj)
 193                 return;
 194
 195         if (!(obj->flags & FLAG_LINK))
 196                 return;
 197
 198         if (!(obj->flags & FLAG_CHECKED)) {
 199                 unsigned long size;
 200                 int type = sha1_object_info(obj->sha1, &size);
 201                 if (type != obj->type || type <= 0)
 202                         die(_("object of unexpected type"));
 203                 obj->flags |= FLAG_CHECKED;
 204                 return;
 205         }
 206 }
 207
 208 static void check_objects(void)
 209 {
 210         unsigned i, max;
 211
 212         max = get_max_object_index();
 213         for (i = 0; i < max; i++)
 214                 check_object(get_indexed_object(i));
 215 }
 216
 217
 218 /* Discard current buffer used content. */
 219 static void flush(void)
 220 {
 221         if (input_offset) {
 222                 if (output_fd >= 0)
 223                         write_or_die(output_fd, input_buffer, input_offset);
 224                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 225                 memmove(input_buffer, input_buffer + input_offset, input_len);
 226                 input_offset = 0;
 227         }
 228 }
 229
 230 /*
 231  * Make sure at least "min" bytes are available in the buffer, and
 232  * return the pointer to the buffer.
 233  */
 234 static void *fill(int min)
 235 {
 236         if (min <= input_len)
 237                 return input_buffer + input_offset;
 238         if (min > sizeof(input_buffer))
 239                 die(Q_("cannot fill %d byte",
 240                        "cannot fill %d bytes",
 241                        min),
 242                     min);
 243         flush();
 244         do {
 245                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 246                                 sizeof(input_buffer) - input_len);
 247                 if (ret <= 0) {
 248                         if (!ret)
 249                                 die(_("early EOF"));
 250                         die_errno(_("read error on input"));
 251                 }
 252                 input_len += ret;
 253                 if (from_stdin)
 254                         display_throughput(progress, consumed_bytes + input_len);
 255         } while (input_len < min);
 256         return input_buffer;
 257 }
 258
 259 static void use(int bytes)
 260 {
 261         if (bytes > input_len)
 262                 die(_("used more bytes than were available"));
 263         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 264         input_len -= bytes;
 265         input_offset += bytes;
 266
 267         /* make sure off_t is sufficiently large not to wrap */
 268         if (signed_add_overflows(consumed_bytes, bytes))
 269                 die(_("pack too large for current definition of off_t"));
 270         consumed_bytes += bytes;
 271 }
 272
 273 static const char *open_pack_file(const char *pack_name)
 274 {
 275         if (from_stdin) {
 276                 input_fd = 0;
 277                 if (!pack_name) {
 278                         static char tmp_file[PATH_MAX];
 279                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 280                                                 "pack/tmp_pack_XXXXXX");
 281                         pack_name = xstrdup(tmp_file);
 282                 } else
 283                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 284                 if (output_fd < 0)
 285                         die_errno(_("unable to create '%s'"), pack_name);
 286                 pack_fd = output_fd;
 287         } else {
 288                 input_fd = open(pack_name, O_RDONLY);
 289                 if (input_fd < 0)
 290                         die_errno(_("cannot open packfile '%s'"), pack_name);
 291                 output_fd = -1;
 292                 pack_fd = input_fd;
 293         }
 294         git_SHA1_Init(&input_ctx);
 295         return pack_name;
 296 }
 297
 298 static void parse_pack_header(void)
 299 {
 300         struct pack_header *hdr = fill(sizeof(struct pack_header));
 301
 302         /* Header consistency check */
 303         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 304                 die(_("pack signature mismatch"));
 305         if (!pack_version_ok(hdr->hdr_version))
 306                 die(_("pack version %"PRIu32" unsupported"),
 307                         ntohl(hdr->hdr_version));
 308
 309         nr_objects = ntohl(hdr->hdr_entries);
 310         use(sizeof(struct pack_header));
 311 }
 312
 313 static NORETURN void bad_object(unsigned long offset, const char *format,
 314                        ...) __attribute__((format (printf, 2, 3)));
 315
 316 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 317 {
 318         va_list params;
 319         char buf[1024];
 320
 321         va_start(params, format);
 322         vsnprintf(buf, sizeof(buf), format, params);
 323         va_end(params);
 324         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 325 }
 326
 327 static inline struct thread_local *get_thread_data(void)
 328 {
 329 #ifndef NO_PTHREADS
 330         if (threads_active)
 331                 return pthread_getspecific(key);
 332         assert(!threads_active &&
 333                "This should only be reached when all threads are gone");
 334 #endif
 335         return &nothread_data;
 336 }
 337
 338 #ifndef NO_PTHREADS
 339 static void set_thread_data(struct thread_local *data)
 340 {
 341         if (threads_active)
 342                 pthread_setspecific(key, data);
 343 }
 344 #endif
 345
 346 static struct base_data *alloc_base_data(void)
 347 {
 348         struct base_data *base = xmalloc(sizeof(struct base_data));
 349         memset(base, 0, sizeof(*base));
 350         base->ref_last = -1;
 351         base->ofs_last = -1;
 352         return base;
 353 }
 354
 355 static void free_base_data(struct base_data *c)
 356 {
 357         if (c->data) {
 358                 free(c->data);
 359                 c->data = NULL;
 360                 get_thread_data()->base_cache_used -= c->size;
 361         }
 362 }
 363
 364 static void prune_base_data(struct base_data *retain)
 365 {
 366         struct base_data *b;
 367         struct thread_local *data = get_thread_data();
 368         for (b = data->base_cache;
 369              data->base_cache_used > delta_base_cache_limit && b;
 370              b = b->child) {
 371                 if (b->data && b != retain)
 372                         free_base_data(b);
 373         }
 374 }
 375
 376 static void link_base_data(struct base_data *base, struct base_data *c)
 377 {
 378         if (base)
 379                 base->child = c;
 380         else
 381                 get_thread_data()->base_cache = c;
 382
 383         c->base = base;
 384         c->child = NULL;
 385         if (c->data)
 386                 get_thread_data()->base_cache_used += c->size;
 387         prune_base_data(c);
 388 }
 389
 390 static void unlink_base_data(struct base_data *c)
 391 {
 392         struct base_data *base = c->base;
 393         if (base)
 394                 base->child = NULL;
 395         else
 396                 get_thread_data()->base_cache = NULL;
 397         free_base_data(c);
 398 }
 399
 400 static int is_delta_type(enum object_type type)
 401 {
 402         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 403 }
 404
 405 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 406                                enum object_type type, unsigned char *sha1)
 407 {
 408         static char fixed_buf[8192];
 409         int status;
 410         git_zstream stream;
 411         void *buf;
 412         git_SHA_CTX c;
 413         char hdr[32];
 414         int hdrlen;
 415
 416         if (!is_delta_type(type)) {
 417                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 418                 git_SHA1_Init(&c);
 419                 git_SHA1_Update(&c, hdr, hdrlen);
 420         } else
 421                 sha1 = NULL;
 422         if (type == OBJ_BLOB && size > big_file_threshold)
 423                 buf = fixed_buf;
 424         else
 425                 buf = xmalloc(size);
 426
 427         memset(&stream, 0, sizeof(stream));
 428         git_inflate_init(&stream);
 429         stream.next_out = buf;
 430         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 431
 432         do {
 433                 unsigned char *last_out = stream.next_out;
 434                 stream.next_in = fill(1);
 435                 stream.avail_in = input_len;
 436                 status = git_inflate(&stream, 0);
 437                 use(input_len - stream.avail_in);
 438                 if (sha1)
 439                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 440                 if (buf == fixed_buf) {
 441                         stream.next_out = buf;
 442                         stream.avail_out = sizeof(fixed_buf);
 443                 }
 444         } while (status == Z_OK);
 445         if (stream.total_out != size || status != Z_STREAM_END)
 446                 bad_object(offset, _("inflate returned %d"), status);
 447         git_inflate_end(&stream);
 448         if (sha1)
 449                 git_SHA1_Final(sha1, &c);
 450         return buf == fixed_buf ? NULL : buf;
 451 }
 452
 453 static void *unpack_raw_entry(struct object_entry *obj,
 454                               union delta_base *delta_base,
 455                               unsigned char *sha1)
 456 {
 457         unsigned char *p;
 458         unsigned long size, c;
 459         off_t base_offset;
 460         unsigned shift;
 461         void *data;
 462
 463         obj->idx.offset = consumed_bytes;
 464         input_crc32 = crc32(0, NULL, 0);
 465
 466         p = fill(1);
 467         c = *p;
 468         use(1);
 469         obj->type = (c >> 4) & 7;
 470         size = (c & 15);
 471         shift = 4;
 472         while (c & 0x80) {
 473                 p = fill(1);
 474                 c = *p;
 475                 use(1);
 476                 size += (c & 0x7f) << shift;
 477                 shift += 7;
 478         }
 479         obj->size = size;
 480
 481         switch (obj->type) {
 482         case OBJ_REF_DELTA:
 483                 hashcpy(delta_base->sha1, fill(20));
 484                 use(20);
 485                 break;
 486         case OBJ_OFS_DELTA:
 487                 memset(delta_base, 0, sizeof(*delta_base));
 488                 p = fill(1);
 489                 c = *p;
 490                 use(1);
 491                 base_offset = c & 127;
 492                 while (c & 128) {
 493                         base_offset += 1;
 494                         if (!base_offset || MSB(base_offset, 7))
 495                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 496                         p = fill(1);
 497                         c = *p;
 498                         use(1);
 499                         base_offset = (base_offset << 7) + (c & 127);
 500                 }
 501                 delta_base->offset = obj->idx.offset - base_offset;
 502                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 503                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 504                 break;
 505         case OBJ_COMMIT:
 506         case OBJ_TREE:
 507         case OBJ_BLOB:
 508         case OBJ_TAG:
 509                 break;
 510         default:
 511                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 512         }
 513         obj->hdr_size = consumed_bytes - obj->idx.offset;
 514
 515         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 516         obj->idx.crc32 = input_crc32;
 517         return data;
 518 }
 519
 520 static void *unpack_data(struct object_entry *obj,
 521                          int (*consume)(const unsigned char *, unsigned long, void *),
 522                          void *cb_data)
 523 {
 524         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 525         unsigned long len = obj[1].idx.offset - from;
 526         unsigned char *data, *inbuf;
 527         git_zstream stream;
 528         int status;
 529
 530         data = xmalloc(consume ? 64*1024 : obj->size);
 531         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 532
 533         memset(&stream, 0, sizeof(stream));
 534         git_inflate_init(&stream);
 535         stream.next_out = data;
 536         stream.avail_out = consume ? 64*1024 : obj->size;
 537
 538         do {
 539                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 540                 n = pread(pack_fd, inbuf, n, from);
 541                 if (n < 0)
 542                         die_errno(_("cannot pread pack file"));
 543                 if (!n)
 544                         die(Q_("premature end of pack file, %lu byte missing",
 545                                "premature end of pack file, %lu bytes missing",
 546                                len),
 547                             len);
 548                 from += n;
 549                 len -= n;
 550                 stream.next_in = inbuf;
 551                 stream.avail_in = n;
 552                 if (!consume)
 553                         status = git_inflate(&stream, 0);
 554                 else {
 555                         do {
 556                                 status = git_inflate(&stream, 0);
 557                                 if (consume(data, stream.next_out - data, cb_data)) {
 558                                         free(inbuf);
 559                                         free(data);
 560                                         return NULL;
 561                                 }
 562                                 stream.next_out = data;
 563                                 stream.avail_out = 64*1024;
 564                         } while (status == Z_OK && stream.avail_in);
 565                 }
 566         } while (len && status == Z_OK && !stream.avail_in);
 567
 568         /* This has been inflated OK when first encountered, so... */
 569         if (status != Z_STREAM_END || stream.total_out != obj->size)
 570                 die(_("serious inflate inconsistency"));
 571
 572         git_inflate_end(&stream);
 573         free(inbuf);
 574         if (consume) {
 575                 free(data);
 576                 data = NULL;
 577         }
 578         return data;
 579 }
 580
 581 static void *get_data_from_pack(struct object_entry *obj)
 582 {
 583         return unpack_data(obj, NULL, NULL);
 584 }
 585
 586 static int compare_delta_bases(const union delta_base *base1,
 587                                const union delta_base *base2,
 588                                enum object_type type1,
 589                                enum object_type type2)
 590 {
 591         int cmp = type1 - type2;
 592         if (cmp)
 593                 return cmp;
 594         return memcmp(base1, base2, UNION_BASE_SZ);
 595 }
 596
 597 static int find_delta(const union delta_base *base, enum object_type type)
 598 {
 599         int first = 0, last = nr_deltas;
 600
 601         while (first < last) {
 602                 int next = (first + last) / 2;
 603                 struct delta_entry *delta = &deltas[next];
 604                 int cmp;
 605
 606                 cmp = compare_delta_bases(base, &delta->base,
 607                                           type, objects[delta->obj_no].type);
 608                 if (!cmp)
 609                         return next;
 610                 if (cmp < 0) {
 611                         last = next;
 612                         continue;
 613                 }
 614                 first = next+1;
 615         }
 616         return -first-1;
 617 }
 618
 619 static void find_delta_children(const union delta_base *base,
 620                                 int *first_index, int *last_index,
 621                                 enum object_type type)
 622 {
 623         int first = find_delta(base, type);
 624         int last = first;
 625         int end = nr_deltas - 1;
 626
 627         if (first < 0) {
 628                 *first_index = 0;
 629                 *last_index = -1;
 630                 return;
 631         }
 632         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 633                 --first;
 634         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 635                 ++last;
 636         *first_index = first;
 637         *last_index = last;
 638 }
 639
 640 struct compare_data {
 641         struct object_entry *entry;
 642         struct git_istream *st;
 643         unsigned char *buf;
 644         unsigned long buf_size;
 645 };
 646
 647 static int compare_objects(const unsigned char *buf, unsigned long size,
 648                            void *cb_data)
 649 {
 650         struct compare_data *data = cb_data;
 651
 652         if (data->buf_size < size) {
 653                 free(data->buf);
 654                 data->buf = xmalloc(size);
 655                 data->buf_size = size;
 656         }
 657
 658         while (size) {
 659                 ssize_t len = read_istream(data->st, data->buf, size);
 660                 if (len == 0)
 661                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 662                             sha1_to_hex(data->entry->idx.sha1));
 663                 if (len < 0)
 664                         die(_("unable to read %s"),
 665                             sha1_to_hex(data->entry->idx.sha1));
 666                 if (memcmp(buf, data->buf, len))
 667                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 668                             sha1_to_hex(data->entry->idx.sha1));
 669                 size -= len;
 670                 buf += len;
 671         }
 672         return 0;
 673 }
 674
 675 static int check_collison(struct object_entry *entry)
 676 {
 677         struct compare_data data;
 678         enum object_type type;
 679         unsigned long size;
 680
 681         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 682                 return -1;
 683
 684         memset(&data, 0, sizeof(data));
 685         data.entry = entry;
 686         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 687         if (!data.st)
 688                 return -1;
 689         if (size != entry->size || type != entry->type)
 690                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 691                     sha1_to_hex(entry->idx.sha1));
 692         unpack_data(entry, compare_objects, &data);
 693         close_istream(data.st);
 694         free(data.buf);
 695         return 0;
 696 }
 697
 698 static void sha1_object(const void *data, struct object_entry *obj_entry,
 699                         unsigned long size, enum object_type type,
 700                         const unsigned char *sha1)
 701 {
 702         void *new_data = NULL;
 703         int collision_test_needed;
 704
 705         assert(data || obj_entry);
 706
 707         read_lock();
 708         collision_test_needed = has_sha1_file(sha1);
 709         read_unlock();
 710
 711         if (collision_test_needed && !data) {
 712                 read_lock();
 713                 if (!check_collison(obj_entry))
 714                         collision_test_needed = 0;
 715                 read_unlock();
 716         }
 717         if (collision_test_needed) {
 718                 void *has_data;
 719                 enum object_type has_type;
 720                 unsigned long has_size;
 721                 read_lock();
 722                 has_type = sha1_object_info(sha1, &has_size);
 723                 if (has_type != type || has_size != size)
 724                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 725                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 726                 read_unlock();
 727                 if (!data)
 728                         data = new_data = get_data_from_pack(obj_entry);
 729                 if (!has_data)
 730                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 731                 if (size != has_size || type != has_type ||
 732                     memcmp(data, has_data, size) != 0)
 733                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 734                 free(has_data);
 735         }
 736
 737         if (strict) {
 738                 read_lock();
 739                 if (type == OBJ_BLOB) {
 740                         struct blob *blob = lookup_blob(sha1);
 741                         if (blob)
 742                                 blob->object.flags |= FLAG_CHECKED;
 743                         else
 744                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 745                 } else {
 746                         struct object *obj;
 747                         int eaten;
 748                         void *buf = (void *) data;
 749
 750                         if (!buf)
 751                                 buf = new_data = get_data_from_pack(obj_entry);
 752
 753                         /*
 754                          * we do not need to free the memory here, as the
 755                          * buf is deleted by the caller.
 756                          */
 757                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 758                         if (!obj)
 759                                 die(_("invalid %s"), typename(type));
 760                         if (fsck_object(obj, 1, fsck_error_function))
 761                                 die(_("Error in object"));
 762                         if (fsck_walk(obj, mark_link, NULL))
 763                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 764
 765                         if (obj->type == OBJ_TREE) {
 766                                 struct tree *item = (struct tree *) obj;
 767                                 item->buffer = NULL;
 768                         }
 769                         if (obj->type == OBJ_COMMIT) {
 770                                 struct commit *commit = (struct commit *) obj;
 771                                 commit->buffer = NULL;
 772                         }
 773                         obj->flags |= FLAG_CHECKED;
 774                 }
 775                 read_unlock();
 776         }
 777
 778         free(new_data);
 779 }
 780
 781 /*
 782  * This function is part of find_unresolved_deltas(). There are two
 783  * walkers going in the opposite ways.
 784  *
 785  * The first one in find_unresolved_deltas() traverses down from
 786  * parent node to children, deflating nodes along the way. However,
 787  * memory for deflated nodes is limited by delta_base_cache_limit, so
 788  * at some point parent node's deflated content may be freed.
 789  *
 790  * The second walker is this function, which goes from current node up
 791  * to top parent if necessary to deflate the node. In normal
 792  * situation, its parent node would be already deflated, so it just
 793  * needs to apply delta.
 794  *
 795  * In the worst case scenario, parent node is no longer deflated because
 796  * we're running out of delta_base_cache_limit; we need to re-deflate
 797  * parents, possibly up to the top base.
 798  *
 799  * All deflated objects here are subject to be freed if we exceed
 800  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 801  * just need to make sure the last node is not freed.
 802  */
 803 static void *get_base_data(struct base_data *c)
 804 {
 805         if (!c->data) {
 806                 struct object_entry *obj = c->obj;
 807                 struct base_data **delta = NULL;
 808                 int delta_nr = 0, delta_alloc = 0;
 809
 810                 while (is_delta_type(c->obj->type) && !c->data) {
 811                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 812                         delta[delta_nr++] = c;
 813                         c = c->base;
 814                 }
 815                 if (!delta_nr) {
 816                         c->data = get_data_from_pack(obj);
 817                         c->size = obj->size;
 818                         get_thread_data()->base_cache_used += c->size;
 819                         prune_base_data(c);
 820                 }
 821                 for (; delta_nr > 0; delta_nr--) {
 822                         void *base, *raw;
 823                         c = delta[delta_nr - 1];
 824                         obj = c->obj;
 825                         base = get_base_data(c->base);
 826                         raw = get_data_from_pack(obj);
 827                         c->data = patch_delta(
 828                                 base, c->base->size,
 829                                 raw, obj->size,
 830                                 &c->size);
 831                         free(raw);
 832                         if (!c->data)
 833                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 834                         get_thread_data()->base_cache_used += c->size;
 835                         prune_base_data(c);
 836                 }
 837                 free(delta);
 838         }
 839         return c->data;
 840 }
 841
 842 static void resolve_delta(struct object_entry *delta_obj,
 843                           struct base_data *base, struct base_data *result)
 844 {
 845         void *base_data, *delta_data;
 846
 847         delta_obj->real_type = base->obj->real_type;
 848         if (show_stat) {
 849                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 850                 deepest_delta_lock();
 851                 if (deepest_delta < delta_obj->delta_depth)
 852                         deepest_delta = delta_obj->delta_depth;
 853                 deepest_delta_unlock();
 854         }
 855         delta_obj->base_object_no = base->obj - objects;
 856         delta_data = get_data_from_pack(delta_obj);
 857         base_data = get_base_data(base);
 858         result->obj = delta_obj;
 859         result->data = patch_delta(base_data, base->size,
 860                                    delta_data, delta_obj->size, &result->size);
 861         free(delta_data);
 862         if (!result->data)
 863                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 864         hash_sha1_file(result->data, result->size,
 865                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 866         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 867                     delta_obj->idx.sha1);
 868         counter_lock();
 869         nr_resolved_deltas++;
 870         counter_unlock();
 871 }
 872
 873 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 874                                                   struct base_data *prev_base)
 875 {
 876         if (base->ref_last == -1 && base->ofs_last == -1) {
 877                 union delta_base base_spec;
 878
 879                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 880                 find_delta_children(&base_spec,
 881                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 882
 883                 memset(&base_spec, 0, sizeof(base_spec));
 884                 base_spec.offset = base->obj->idx.offset;
 885                 find_delta_children(&base_spec,
 886                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 887
 888                 if (base->ref_last == -1 && base->ofs_last == -1) {
 889                         free(base->data);
 890                         return NULL;
 891                 }
 892
 893                 link_base_data(prev_base, base);
 894         }
 895
 896         if (base->ref_first <= base->ref_last) {
 897                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 898                 struct base_data *result = alloc_base_data();
 899
 900                 assert(child->real_type == OBJ_REF_DELTA);
 901                 resolve_delta(child, base, result);
 902                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 903                         free_base_data(base);
 904
 905                 base->ref_first++;
 906                 return result;
 907         }
 908
 909         if (base->ofs_first <= base->ofs_last) {
 910                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 911                 struct base_data *result = alloc_base_data();
 912
 913                 assert(child->real_type == OBJ_OFS_DELTA);
 914                 resolve_delta(child, base, result);
 915                 if (base->ofs_first == base->ofs_last)
 916                         free_base_data(base);
 917
 918                 base->ofs_first++;
 919                 return result;
 920         }
 921
 922         unlink_base_data(base);
 923         return NULL;
 924 }
 925
 926 static void find_unresolved_deltas(struct base_data *base)
 927 {
 928         struct base_data *new_base, *prev_base = NULL;
 929         for (;;) {
 930                 new_base = find_unresolved_deltas_1(base, prev_base);
 931
 932                 if (new_base) {
 933                         prev_base = base;
 934                         base = new_base;
 935                 } else {
 936                         free(base);
 937                         base = prev_base;
 938                         if (!base)
 939                                 return;
 940                         prev_base = base->base;
 941                 }
 942         }
 943 }
 944
 945 static int compare_delta_entry(const void *a, const void *b)
 946 {
 947         const struct delta_entry *delta_a = a;
 948         const struct delta_entry *delta_b = b;
 949
 950         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 951         return compare_delta_bases(&delta_a->base, &delta_b->base,
 952                                    objects[delta_a->obj_no].type,
 953                                    objects[delta_b->obj_no].type);
 954 }
 955
 956 static void resolve_base(struct object_entry *obj)
 957 {
 958         struct base_data *base_obj = alloc_base_data();
 959         base_obj->obj = obj;
 960         base_obj->data = NULL;
 961         find_unresolved_deltas(base_obj);
 962 }
 963
 964 #ifndef NO_PTHREADS
 965 static void *threaded_second_pass(void *data)
 966 {
 967         set_thread_data(data);
 968         for (;;) {
 969                 int i;
 970                 counter_lock();
 971                 display_progress(progress, nr_resolved_deltas);
 972                 counter_unlock();
 973                 work_lock();
 974                 while (nr_dispatched < nr_objects &&
 975                        is_delta_type(objects[nr_dispatched].type))
 976                         nr_dispatched++;
 977                 if (nr_dispatched >= nr_objects) {
 978                         work_unlock();
 979                         break;
 980                 }
 981                 i = nr_dispatched++;
 982                 work_unlock();
 983
 984                 resolve_base(&objects[i]);
 985         }
 986         return NULL;
 987 }
 988 #endif
 989
 990 /*
 991  * First pass:
 992  * - find locations of all objects;
 993  * - calculate SHA1 of all non-delta objects;
 994  * - remember base (SHA1 or offset) for all deltas.
 995  */
 996 static void parse_pack_objects(unsigned char *sha1)
 997 {
 998         int i, nr_delays = 0;
 999         struct delta_entry *delta = deltas;
1000         struct stat st;
1001
1002         if (verbose)
1003                 progress = start_progress(
1004                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1005                                 nr_objects);
1006         for (i = 0; i < nr_objects; i++) {
1007                 struct object_entry *obj = &objects[i];
1008                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1009                 obj->real_type = obj->type;
1010                 if (is_delta_type(obj->type)) {
1011                         nr_deltas++;
1012                         delta->obj_no = i;
1013                         delta++;
1014                 } else if (!data) {
1015                         /* large blobs, check later */
1016                         obj->real_type = OBJ_BAD;
1017                         nr_delays++;
1018                 } else
1019                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1020                 free(data);
1021                 display_progress(progress, i+1);
1022         }
1023         objects[i].idx.offset = consumed_bytes;
1024         stop_progress(&progress);
1025
1026         /* Check pack integrity */
1027         flush();
1028         git_SHA1_Final(sha1, &input_ctx);
1029         if (hashcmp(fill(20), sha1))
1030                 die(_("pack is corrupted (SHA1 mismatch)"));
1031         use(20);
1032
1033         /* If input_fd is a file, we should have reached its end now. */
1034         if (fstat(input_fd, &st))
1035                 die_errno(_("cannot fstat packfile"));
1036         if (S_ISREG(st.st_mode) &&
1037                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1038                 die(_("pack has junk at the end"));
1039
1040         for (i = 0; i < nr_objects; i++) {
1041                 struct object_entry *obj = &objects[i];
1042                 if (obj->real_type != OBJ_BAD)
1043                         continue;
1044                 obj->real_type = obj->type;
1045                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1046                 nr_delays--;
1047         }
1048         if (nr_delays)
1049                 die(_("confusion beyond insanity in parse_pack_objects()"));
1050 }
1051
1052 /*
1053  * Second pass:
1054  * - for all non-delta objects, look if it is used as a base for
1055  *   deltas;
1056  * - if used as a base, uncompress the object and apply all deltas,
1057  *   recursively checking if the resulting object is used as a base
1058  *   for some more deltas.
1059  */
1060 static void resolve_deltas(void)
1061 {
1062         int i;
1063
1064         if (!nr_deltas)
1065                 return;
1066
1067         /* Sort deltas by base SHA1/offset for fast searching */
1068         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1069               compare_delta_entry);
1070
1071         if (verbose)
1072                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1073
1074 #ifndef NO_PTHREADS
1075         nr_dispatched = 0;
1076         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1077                 init_thread();
1078                 for (i = 0; i < nr_threads; i++) {
1079                         int ret = pthread_create(&thread_data[i].thread, NULL,
1080                                                  threaded_second_pass, thread_data + i);
1081                         if (ret)
1082                                 die(_("unable to create thread: %s"),
1083                                     strerror(ret));
1084                 }
1085                 for (i = 0; i < nr_threads; i++)
1086                         pthread_join(thread_data[i].thread, NULL);
1087                 cleanup_thread();
1088                 return;
1089         }
1090 #endif
1091
1092         for (i = 0; i < nr_objects; i++) {
1093                 struct object_entry *obj = &objects[i];
1094
1095                 if (is_delta_type(obj->type))
1096                         continue;
1097                 resolve_base(obj);
1098                 display_progress(progress, nr_resolved_deltas);
1099         }
1100 }
1101
1102 /*
1103  * Third pass:
1104  * - append objects to convert thin pack to full pack if required
1105  * - write the final 20-byte SHA-1
1106  */
1107 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1108 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1109 {
1110         if (nr_deltas == nr_resolved_deltas) {
1111                 stop_progress(&progress);
1112                 /* Flush remaining pack final 20-byte SHA1. */
1113                 flush();
1114                 return;
1115         }
1116
1117         if (fix_thin_pack) {
1118                 struct sha1file *f;
1119                 unsigned char read_sha1[20], tail_sha1[20];
1120                 struct strbuf msg = STRBUF_INIT;
1121                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1122                 int nr_objects_initial = nr_objects;
1123                 if (nr_unresolved <= 0)
1124                         die(_("confusion beyond insanity"));
1125                 objects = xrealloc(objects,
1126                                    (nr_objects + nr_unresolved + 1)
1127                                    * sizeof(*objects));
1128                 memset(objects + nr_objects + 1, 0,
1129                        nr_unresolved * sizeof(*objects));
1130                 f = sha1fd(output_fd, curr_pack);
1131                 fix_unresolved_deltas(f, nr_unresolved);
1132                 strbuf_addf(&msg, _("completed with %d local objects"),
1133                             nr_objects - nr_objects_initial);
1134                 stop_progress_msg(&progress, msg.buf);
1135                 strbuf_release(&msg);
1136                 sha1close(f, tail_sha1, 0);
1137                 hashcpy(read_sha1, pack_sha1);
1138                 fixup_pack_header_footer(output_fd, pack_sha1,
1139                                          curr_pack, nr_objects,
1140                                          read_sha1, consumed_bytes-20);
1141                 if (hashcmp(read_sha1, tail_sha1) != 0)
1142                         die(_("Unexpected tail checksum for %s "
1143                               "(disk corruption?)"), curr_pack);
1144         }
1145         if (nr_deltas != nr_resolved_deltas)
1146                 die(Q_("pack has %d unresolved delta",
1147                        "pack has %d unresolved deltas",
1148                        nr_deltas - nr_resolved_deltas),
1149                     nr_deltas - nr_resolved_deltas);
1150 }
1151
1152 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1153 {
1154         git_zstream stream;
1155         int status;
1156         unsigned char outbuf[4096];
1157
1158         memset(&stream, 0, sizeof(stream));
1159         git_deflate_init(&stream, zlib_compression_level);
1160         stream.next_in = in;
1161         stream.avail_in = size;
1162
1163         do {
1164                 stream.next_out = outbuf;
1165                 stream.avail_out = sizeof(outbuf);
1166                 status = git_deflate(&stream, Z_FINISH);
1167                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1168         } while (status == Z_OK);
1169
1170         if (status != Z_STREAM_END)
1171                 die(_("unable to deflate appended object (%d)"), status);
1172         size = stream.total_out;
1173         git_deflate_end(&stream);
1174         return size;
1175 }
1176
1177 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1178                                const unsigned char *sha1, void *buf,
1179                                unsigned long size, enum object_type type)
1180 {
1181         struct object_entry *obj = &objects[nr_objects++];
1182         unsigned char header[10];
1183         unsigned long s = size;
1184         int n = 0;
1185         unsigned char c = (type << 4) | (s & 15);
1186         s >>= 4;
1187         while (s) {
1188                 header[n++] = c | 0x80;
1189                 c = s & 0x7f;
1190                 s >>= 7;
1191         }
1192         header[n++] = c;
1193         crc32_begin(f);
1194         sha1write(f, header, n);
1195         obj[0].size = size;
1196         obj[0].hdr_size = n;
1197         obj[0].type = type;
1198         obj[0].real_type = type;
1199         obj[1].idx.offset = obj[0].idx.offset + n;
1200         obj[1].idx.offset += write_compressed(f, buf, size);
1201         obj[0].idx.crc32 = crc32_end(f);
1202         sha1flush(f);
1203         hashcpy(obj->idx.sha1, sha1);
1204         return obj;
1205 }
1206
1207 static int delta_pos_compare(const void *_a, const void *_b)
1208 {
1209         struct delta_entry *a = *(struct delta_entry **)_a;
1210         struct delta_entry *b = *(struct delta_entry **)_b;
1211         return a->obj_no - b->obj_no;
1212 }
1213
1214 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1215 {
1216         struct delta_entry **sorted_by_pos;
1217         int i, n = 0;
1218
1219         /*
1220          * Since many unresolved deltas may well be themselves base objects
1221          * for more unresolved deltas, we really want to include the
1222          * smallest number of base objects that would cover as much delta
1223          * as possible by picking the
1224          * trunc deltas first, allowing for other deltas to resolve without
1225          * additional base objects.  Since most base objects are to be found
1226          * before deltas depending on them, a good heuristic is to start
1227          * resolving deltas in the same order as their position in the pack.
1228          */
1229         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1230         for (i = 0; i < nr_deltas; i++) {
1231                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1232                         continue;
1233                 sorted_by_pos[n++] = &deltas[i];
1234         }
1235         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1236
1237         for (i = 0; i < n; i++) {
1238                 struct delta_entry *d = sorted_by_pos[i];
1239                 enum object_type type;
1240                 struct base_data *base_obj = alloc_base_data();
1241
1242                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1243                         continue;
1244                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1245                 if (!base_obj->data)
1246                         continue;
1247
1248                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1249                                 base_obj->size, typename(type)))
1250                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1251                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1252                                         base_obj->data, base_obj->size, type);
1253                 find_unresolved_deltas(base_obj);
1254                 display_progress(progress, nr_resolved_deltas);
1255         }
1256         free(sorted_by_pos);
1257 }
1258
1259 static void final(const char *final_pack_name, const char *curr_pack_name,
1260                   const char *final_index_name, const char *curr_index_name,
1261                   const char *keep_name, const char *keep_msg,
1262                   unsigned char *sha1)
1263 {
1264         const char *report = "pack";
1265         char name[PATH_MAX];
1266         int err;
1267
1268         if (!from_stdin) {
1269                 close(input_fd);
1270         } else {
1271                 fsync_or_die(output_fd, curr_pack_name);
1272                 err = close(output_fd);
1273                 if (err)
1274                         die_errno(_("error while closing pack file"));
1275         }
1276
1277         if (keep_msg) {
1278                 int keep_fd, keep_msg_len = strlen(keep_msg);
1279
1280                 if (!keep_name)
1281                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1282                 else
1283                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1284
1285                 if (keep_fd < 0) {
1286                         if (errno != EEXIST)
1287                                 die_errno(_("cannot write keep file '%s'"),
1288                                           keep_name);
1289                 } else {
1290                         if (keep_msg_len > 0) {
1291                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1292                                 write_or_die(keep_fd, "\n", 1);
1293                         }
1294                         if (close(keep_fd) != 0)
1295                                 die_errno(_("cannot close written keep file '%s'"),
1296                                     keep_name);
1297                         report = "keep";
1298                 }
1299         }
1300
1301         if (final_pack_name != curr_pack_name) {
1302                 if (!final_pack_name) {
1303                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1304                                  get_object_directory(), sha1_to_hex(sha1));
1305                         final_pack_name = name;
1306                 }
1307                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1308                         die(_("cannot store pack file"));
1309         } else if (from_stdin)
1310                 chmod(final_pack_name, 0444);
1311
1312         if (final_index_name != curr_index_name) {
1313                 if (!final_index_name) {
1314                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1315                                  get_object_directory(), sha1_to_hex(sha1));
1316                         final_index_name = name;
1317                 }
1318                 if (move_temp_to_file(curr_index_name, final_index_name))
1319                         die(_("cannot store index file"));
1320         } else
1321                 chmod(final_index_name, 0444);
1322
1323         if (!from_stdin) {
1324                 printf("%s\n", sha1_to_hex(sha1));
1325         } else {
1326                 char buf[48];
1327                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1328                                    report, sha1_to_hex(sha1));
1329                 write_or_die(1, buf, len);
1330
1331                 /*
1332                  * Let's just mimic git-unpack-objects here and write
1333                  * the last part of the input buffer to stdout.
1334                  */
1335                 while (input_len) {
1336                         err = xwrite(1, input_buffer + input_offset, input_len);
1337                         if (err <= 0)
1338                                 break;
1339                         input_len -= err;
1340                         input_offset += err;
1341                 }
1342         }
1343 }
1344
1345 static int git_index_pack_config(const char *k, const char *v, void *cb)
1346 {
1347         struct pack_idx_option *opts = cb;
1348
1349         if (!strcmp(k, "pack.indexversion")) {
1350                 opts->version = git_config_int(k, v);
1351                 if (opts->version > 2)
1352                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1353                 return 0;
1354         }
1355         if (!strcmp(k, "pack.threads")) {
1356                 nr_threads = git_config_int(k, v);
1357                 if (nr_threads < 0)
1358                         die(_("invalid number of threads specified (%d)"),
1359                             nr_threads);
1360 #ifdef NO_PTHREADS
1361                 if (nr_threads != 1)
1362                         warning(_("no threads support, ignoring %s"), k);
1363                 nr_threads = 1;
1364 #endif
1365                 return 0;
1366         }
1367         return git_default_config(k, v, cb);
1368 }
1369
1370 static int cmp_uint32(const void *a_, const void *b_)
1371 {
1372         uint32_t a = *((uint32_t *)a_);
1373         uint32_t b = *((uint32_t *)b_);
1374
1375         return (a < b) ? -1 : (a != b);
1376 }
1377
1378 static void read_v2_anomalous_offsets(struct packed_git *p,
1379                                       struct pack_idx_option *opts)
1380 {
1381         const uint32_t *idx1, *idx2;
1382         uint32_t i;
1383
1384         /* The address of the 4-byte offset table */
1385         idx1 = (((const uint32_t *)p->index_data)
1386                 + 2 /* 8-byte header */
1387                 + 256 /* fan out */
1388                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1389                 + p->num_objects /* CRC32 table */
1390                 );
1391
1392         /* The address of the 8-byte offset table */
1393         idx2 = idx1 + p->num_objects;
1394
1395         for (i = 0; i < p->num_objects; i++) {
1396                 uint32_t off = ntohl(idx1[i]);
1397                 if (!(off & 0x80000000))
1398                         continue;
1399                 off = off & 0x7fffffff;
1400                 if (idx2[off * 2])
1401                         continue;
1402                 /*
1403                  * The real offset is ntohl(idx2[off * 2]) in high 4
1404                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1405                  * octets.  But idx2[off * 2] is Zero!!!
1406                  */
1407                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1408                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1409         }
1410
1411         if (1 < opts->anomaly_nr)
1412                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1413 }
1414
1415 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1416 {
1417         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1418
1419         if (!p)
1420                 die(_("Cannot open existing pack file '%s'"), pack_name);
1421         if (open_pack_index(p))
1422                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1423
1424         /* Read the attributes from the existing idx file */
1425         opts->version = p->index_version;
1426
1427         if (opts->version == 2)
1428                 read_v2_anomalous_offsets(p, opts);
1429
1430         /*
1431          * Get rid of the idx file as we do not need it anymore.
1432          * NEEDSWORK: extract this bit from free_pack_by_name() in
1433          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1434          * know we haven't installed this pack (hence we never have
1435          * read anything from it).
1436          */
1437         close_pack_index(p);
1438         free(p);
1439 }
1440
1441 static void show_pack_info(int stat_only)
1442 {
1443         int i, baseobjects = nr_objects - nr_deltas;
1444         unsigned long *chain_histogram = NULL;
1445
1446         if (deepest_delta)
1447                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1448
1449         for (i = 0; i < nr_objects; i++) {
1450                 struct object_entry *obj = &objects[i];
1451
1452                 if (is_delta_type(obj->type))
1453                         chain_histogram[obj->delta_depth - 1]++;
1454                 if (stat_only)
1455                         continue;
1456                 printf("%s %-6s %lu %lu %"PRIuMAX,
1457                        sha1_to_hex(obj->idx.sha1),
1458                        typename(obj->real_type), obj->size,
1459                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1460                        (uintmax_t)obj->idx.offset);
1461                 if (is_delta_type(obj->type)) {
1462                         struct object_entry *bobj = &objects[obj->base_object_no];
1463                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1464                 }
1465                 putchar('\n');
1466         }
1467
1468         if (baseobjects)
1469                 printf_ln(Q_("non delta: %d object",
1470                              "non delta: %d objects",
1471                              baseobjects),
1472                           baseobjects);
1473         for (i = 0; i < deepest_delta; i++) {
1474                 if (!chain_histogram[i])
1475                         continue;
1476                 printf_ln(Q_("chain length = %d: %lu object",
1477                              "chain length = %d: %lu objects",
1478                              chain_histogram[i]),
1479                           i + 1,
1480                           chain_histogram[i]);
1481         }
1482 }
1483
1484 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1485 {
1486         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1487         const char *curr_pack, *curr_index;
1488         const char *index_name = NULL, *pack_name = NULL;
1489         const char *keep_name = NULL, *keep_msg = NULL;
1490         char *index_name_buf = NULL, *keep_name_buf = NULL;
1491         struct pack_idx_entry **idx_objects;
1492         struct pack_idx_option opts;
1493         unsigned char pack_sha1[20];
1494
1495         if (argc == 2 && !strcmp(argv[1], "-h"))
1496                 usage(index_pack_usage);
1497
1498         read_replace_refs = 0;
1499
1500         reset_pack_idx_option(&opts);
1501         git_config(git_index_pack_config, &opts);
1502         if (prefix && chdir(prefix))
1503                 die(_("Cannot come back to cwd"));
1504
1505         for (i = 1; i < argc; i++) {
1506                 const char *arg = argv[i];
1507
1508                 if (*arg == '-') {
1509                         if (!strcmp(arg, "--stdin")) {
1510                                 from_stdin = 1;
1511                         } else if (!strcmp(arg, "--fix-thin")) {
1512                                 fix_thin_pack = 1;
1513                         } else if (!strcmp(arg, "--strict")) {
1514                                 strict = 1;
1515                         } else if (!strcmp(arg, "--verify")) {
1516                                 verify = 1;
1517                         } else if (!strcmp(arg, "--verify-stat")) {
1518                                 verify = 1;
1519                                 show_stat = 1;
1520                         } else if (!strcmp(arg, "--verify-stat-only")) {
1521                                 verify = 1;
1522                                 show_stat = 1;
1523                                 stat_only = 1;
1524                         } else if (!strcmp(arg, "--keep")) {
1525                                 keep_msg = "";
1526                         } else if (!prefixcmp(arg, "--keep=")) {
1527                                 keep_msg = arg + 7;
1528                         } else if (!prefixcmp(arg, "--threads=")) {
1529                                 char *end;
1530                                 nr_threads = strtoul(arg+10, &end, 0);
1531                                 if (!arg[10] || *end || nr_threads < 0)
1532                                         usage(index_pack_usage);
1533 #ifdef NO_PTHREADS
1534                                 if (nr_threads != 1)
1535                                         warning(_("no threads support, "
1536                                                   "ignoring %s"), arg);
1537                                 nr_threads = 1;
1538 #endif
1539                         } else if (!prefixcmp(arg, "--pack_header=")) {
1540                                 struct pack_header *hdr;
1541                                 char *c;
1542
1543                                 hdr = (struct pack_header *)input_buffer;
1544                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1545                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1546                                 if (*c != ',')
1547                                         die(_("bad %s"), arg);
1548                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1549                                 if (*c)
1550                                         die(_("bad %s"), arg);
1551                                 input_len = sizeof(*hdr);
1552                         } else if (!strcmp(arg, "-v")) {
1553                                 verbose = 1;
1554                         } else if (!strcmp(arg, "-o")) {
1555                                 if (index_name || (i+1) >= argc)
1556                                         usage(index_pack_usage);
1557                                 index_name = argv[++i];
1558                         } else if (!prefixcmp(arg, "--index-version=")) {
1559                                 char *c;
1560                                 opts.version = strtoul(arg + 16, &c, 10);
1561                                 if (opts.version > 2)
1562                                         die(_("bad %s"), arg);
1563                                 if (*c == ',')
1564                                         opts.off32_limit = strtoul(c+1, &c, 0);
1565                                 if (*c || opts.off32_limit & 0x80000000)
1566                                         die(_("bad %s"), arg);
1567                         } else
1568                                 usage(index_pack_usage);
1569                         continue;
1570                 }
1571
1572                 if (pack_name)
1573                         usage(index_pack_usage);
1574                 pack_name = arg;
1575         }
1576
1577         if (!pack_name && !from_stdin)
1578                 usage(index_pack_usage);
1579         if (fix_thin_pack && !from_stdin)
1580                 die(_("--fix-thin cannot be used without --stdin"));
1581         if (!index_name && pack_name) {
1582                 int len = strlen(pack_name);
1583                 if (!has_extension(pack_name, ".pack"))
1584                         die(_("packfile name '%s' does not end with '.pack'"),
1585                             pack_name);
1586                 index_name_buf = xmalloc(len);
1587                 memcpy(index_name_buf, pack_name, len - 5);
1588                 strcpy(index_name_buf + len - 5, ".idx");
1589                 index_name = index_name_buf;
1590         }
1591         if (keep_msg && !keep_name && pack_name) {
1592                 int len = strlen(pack_name);
1593                 if (!has_extension(pack_name, ".pack"))
1594                         die(_("packfile name '%s' does not end with '.pack'"),
1595                             pack_name);
1596                 keep_name_buf = xmalloc(len);
1597                 memcpy(keep_name_buf, pack_name, len - 5);
1598                 strcpy(keep_name_buf + len - 5, ".keep");
1599                 keep_name = keep_name_buf;
1600         }
1601         if (verify) {
1602                 if (!index_name)
1603                         die(_("--verify with no packfile name given"));
1604                 read_idx_option(&opts, index_name);
1605                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1606         }
1607         if (strict)
1608                 opts.flags |= WRITE_IDX_STRICT;
1609
1610 #ifndef NO_PTHREADS
1611         if (!nr_threads) {
1612                 nr_threads = online_cpus();
1613                 /* An experiment showed that more threads does not mean faster */
1614                 if (nr_threads > 3)
1615                         nr_threads = 3;
1616         }
1617 #endif
1618
1619         curr_pack = open_pack_file(pack_name);
1620         parse_pack_header();
1621         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1622         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1623         parse_pack_objects(pack_sha1);
1624         resolve_deltas();
1625         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1626         free(deltas);
1627         if (strict)
1628                 check_objects();
1629
1630         if (show_stat)
1631                 show_pack_info(stat_only);
1632
1633         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1634         for (i = 0; i < nr_objects; i++)
1635                 idx_objects[i] = &objects[i].idx;
1636         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1637         free(idx_objects);
1638
1639         if (!verify)
1640                 final(pack_name, curr_pack,
1641                       index_name, curr_index,
1642                       keep_name, keep_msg,
1643                       pack_sha1);
1644         else
1645                 close(input_fd);
1646         free(objects);
1647         free(index_name_buf);
1648         free(keep_name_buf);
1649         if (pack_name == NULL)
1650                 free((void *) curr_pack);
1651         if (index_name == NULL)
1652                 free((void *) curr_index);
1653
1654         return 0;
1655 }