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 do_fsck_object;
  81 static int verbose;
  82 static int show_stat;
  83 static int check_self_contained_and_connected;
  84
  85 static struct progress *progress;
  86
  87 /* We always read in 4kB chunks. */
  88 static unsigned char input_buffer[4096];
  89 static unsigned int input_offset, input_len;
  90 static off_t consumed_bytes;
  91 static unsigned deepest_delta;
  92 static git_SHA_CTX input_ctx;
  93 static uint32_t input_crc32;
  94 static int input_fd, output_fd, pack_fd;
  95
  96 #ifndef NO_PTHREADS
  97
  98 static struct thread_local *thread_data;
  99 static int nr_dispatched;
 100 static int threads_active;
 101
 102 static pthread_mutex_t read_mutex;
 103 #define read_lock()             lock_mutex(&read_mutex)
 104 #define read_unlock()           unlock_mutex(&read_mutex)
 105
 106 static pthread_mutex_t counter_mutex;
 107 #define counter_lock()          lock_mutex(&counter_mutex)
 108 #define counter_unlock()        unlock_mutex(&counter_mutex)
 109
 110 static pthread_mutex_t work_mutex;
 111 #define work_lock()             lock_mutex(&work_mutex)
 112 #define work_unlock()           unlock_mutex(&work_mutex)
 113
 114 static pthread_mutex_t deepest_delta_mutex;
 115 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 116 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 117
 118 static pthread_key_t key;
 119
 120 static inline void lock_mutex(pthread_mutex_t *mutex)
 121 {
 122         if (threads_active)
 123                 pthread_mutex_lock(mutex);
 124 }
 125
 126 static inline void unlock_mutex(pthread_mutex_t *mutex)
 127 {
 128         if (threads_active)
 129                 pthread_mutex_unlock(mutex);
 130 }
 131
 132 /*
 133  * Mutex and conditional variable can't be statically-initialized on Windows.
 134  */
 135 static void init_thread(void)
 136 {
 137         init_recursive_mutex(&read_mutex);
 138         pthread_mutex_init(&counter_mutex, NULL);
 139         pthread_mutex_init(&work_mutex, NULL);
 140         if (show_stat)
 141                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 142         pthread_key_create(&key, NULL);
 143         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 144         threads_active = 1;
 145 }
 146
 147 static void cleanup_thread(void)
 148 {
 149         if (!threads_active)
 150                 return;
 151         threads_active = 0;
 152         pthread_mutex_destroy(&read_mutex);
 153         pthread_mutex_destroy(&counter_mutex);
 154         pthread_mutex_destroy(&work_mutex);
 155         if (show_stat)
 156                 pthread_mutex_destroy(&deepest_delta_mutex);
 157         pthread_key_delete(key);
 158         free(thread_data);
 159 }
 160
 161 #else
 162
 163 #define read_lock()
 164 #define read_unlock()
 165
 166 #define counter_lock()
 167 #define counter_unlock()
 168
 169 #define work_lock()
 170 #define work_unlock()
 171
 172 #define deepest_delta_lock()
 173 #define deepest_delta_unlock()
 174
 175 #endif
 176
 177
 178 static int mark_link(struct object *obj, int type, void *data)
 179 {
 180         if (!obj)
 181                 return -1;
 182
 183         if (type != OBJ_ANY && obj->type != type)
 184                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 185
 186         obj->flags |= FLAG_LINK;
 187         return 0;
 188 }
 189
 190 /* The content of each linked object must have been checked
 191    or it must be already present in the object database */
 192 static unsigned check_object(struct object *obj)
 193 {
 194         if (!obj)
 195                 return 0;
 196
 197         if (!(obj->flags & FLAG_LINK))
 198                 return 0;
 199
 200         if (!(obj->flags & FLAG_CHECKED)) {
 201                 unsigned long size;
 202                 int type = sha1_object_info(obj->sha1, &size);
 203                 if (type != obj->type || type <= 0)
 204                         die(_("object of unexpected type"));
 205                 obj->flags |= FLAG_CHECKED;
 206                 return 1;
 207         }
 208
 209         return 0;
 210 }
 211
 212 static unsigned check_objects(void)
 213 {
 214         unsigned i, max, foreign_nr = 0;
 215
 216         max = get_max_object_index();
 217         for (i = 0; i < max; i++)
 218                 foreign_nr += check_object(get_indexed_object(i));
 219         return foreign_nr;
 220 }
 221
 222
 223 /* Discard current buffer used content. */
 224 static void flush(void)
 225 {
 226         if (input_offset) {
 227                 if (output_fd >= 0)
 228                         write_or_die(output_fd, input_buffer, input_offset);
 229                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 230                 memmove(input_buffer, input_buffer + input_offset, input_len);
 231                 input_offset = 0;
 232         }
 233 }
 234
 235 /*
 236  * Make sure at least "min" bytes are available in the buffer, and
 237  * return the pointer to the buffer.
 238  */
 239 static void *fill(int min)
 240 {
 241         if (min <= input_len)
 242                 return input_buffer + input_offset;
 243         if (min > sizeof(input_buffer))
 244                 die(Q_("cannot fill %d byte",
 245                        "cannot fill %d bytes",
 246                        min),
 247                     min);
 248         flush();
 249         do {
 250                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 251                                 sizeof(input_buffer) - input_len);
 252                 if (ret <= 0) {
 253                         if (!ret)
 254                                 die(_("early EOF"));
 255                         die_errno(_("read error on input"));
 256                 }
 257                 input_len += ret;
 258                 if (from_stdin)
 259                         display_throughput(progress, consumed_bytes + input_len);
 260         } while (input_len < min);
 261         return input_buffer;
 262 }
 263
 264 static void use(int bytes)
 265 {
 266         if (bytes > input_len)
 267                 die(_("used more bytes than were available"));
 268         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 269         input_len -= bytes;
 270         input_offset += bytes;
 271
 272         /* make sure off_t is sufficiently large not to wrap */
 273         if (signed_add_overflows(consumed_bytes, bytes))
 274                 die(_("pack too large for current definition of off_t"));
 275         consumed_bytes += bytes;
 276 }
 277
 278 static const char *open_pack_file(const char *pack_name)
 279 {
 280         if (from_stdin) {
 281                 input_fd = 0;
 282                 if (!pack_name) {
 283                         static char tmp_file[PATH_MAX];
 284                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 285                                                 "pack/tmp_pack_XXXXXX");
 286                         pack_name = xstrdup(tmp_file);
 287                 } else
 288                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 289                 if (output_fd < 0)
 290                         die_errno(_("unable to create '%s'"), pack_name);
 291                 pack_fd = output_fd;
 292         } else {
 293                 input_fd = open(pack_name, O_RDONLY);
 294                 if (input_fd < 0)
 295                         die_errno(_("cannot open packfile '%s'"), pack_name);
 296                 output_fd = -1;
 297                 pack_fd = input_fd;
 298         }
 299         git_SHA1_Init(&input_ctx);
 300         return pack_name;
 301 }
 302
 303 static void parse_pack_header(void)
 304 {
 305         struct pack_header *hdr = fill(sizeof(struct pack_header));
 306
 307         /* Header consistency check */
 308         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 309                 die(_("pack signature mismatch"));
 310         if (!pack_version_ok(hdr->hdr_version))
 311                 die(_("pack version %"PRIu32" unsupported"),
 312                         ntohl(hdr->hdr_version));
 313
 314         nr_objects = ntohl(hdr->hdr_entries);
 315         use(sizeof(struct pack_header));
 316 }
 317
 318 static NORETURN void bad_object(unsigned long offset, const char *format,
 319                        ...) __attribute__((format (printf, 2, 3)));
 320
 321 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 322 {
 323         va_list params;
 324         char buf[1024];
 325
 326         va_start(params, format);
 327         vsnprintf(buf, sizeof(buf), format, params);
 328         va_end(params);
 329         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 330 }
 331
 332 static inline struct thread_local *get_thread_data(void)
 333 {
 334 #ifndef NO_PTHREADS
 335         if (threads_active)
 336                 return pthread_getspecific(key);
 337         assert(!threads_active &&
 338                "This should only be reached when all threads are gone");
 339 #endif
 340         return &nothread_data;
 341 }
 342
 343 #ifndef NO_PTHREADS
 344 static void set_thread_data(struct thread_local *data)
 345 {
 346         if (threads_active)
 347                 pthread_setspecific(key, data);
 348 }
 349 #endif
 350
 351 static struct base_data *alloc_base_data(void)
 352 {
 353         struct base_data *base = xmalloc(sizeof(struct base_data));
 354         memset(base, 0, sizeof(*base));
 355         base->ref_last = -1;
 356         base->ofs_last = -1;
 357         return base;
 358 }
 359
 360 static void free_base_data(struct base_data *c)
 361 {
 362         if (c->data) {
 363                 free(c->data);
 364                 c->data = NULL;
 365                 get_thread_data()->base_cache_used -= c->size;
 366         }
 367 }
 368
 369 static void prune_base_data(struct base_data *retain)
 370 {
 371         struct base_data *b;
 372         struct thread_local *data = get_thread_data();
 373         for (b = data->base_cache;
 374              data->base_cache_used > delta_base_cache_limit && b;
 375              b = b->child) {
 376                 if (b->data && b != retain)
 377                         free_base_data(b);
 378         }
 379 }
 380
 381 static void link_base_data(struct base_data *base, struct base_data *c)
 382 {
 383         if (base)
 384                 base->child = c;
 385         else
 386                 get_thread_data()->base_cache = c;
 387
 388         c->base = base;
 389         c->child = NULL;
 390         if (c->data)
 391                 get_thread_data()->base_cache_used += c->size;
 392         prune_base_data(c);
 393 }
 394
 395 static void unlink_base_data(struct base_data *c)
 396 {
 397         struct base_data *base = c->base;
 398         if (base)
 399                 base->child = NULL;
 400         else
 401                 get_thread_data()->base_cache = NULL;
 402         free_base_data(c);
 403 }
 404
 405 static int is_delta_type(enum object_type type)
 406 {
 407         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 408 }
 409
 410 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 411                                enum object_type type, unsigned char *sha1)
 412 {
 413         static char fixed_buf[8192];
 414         int status;
 415         git_zstream stream;
 416         void *buf;
 417         git_SHA_CTX c;
 418         char hdr[32];
 419         int hdrlen;
 420
 421         if (!is_delta_type(type)) {
 422                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 423                 git_SHA1_Init(&c);
 424                 git_SHA1_Update(&c, hdr, hdrlen);
 425         } else
 426                 sha1 = NULL;
 427         if (type == OBJ_BLOB && size > big_file_threshold)
 428                 buf = fixed_buf;
 429         else
 430                 buf = xmalloc(size);
 431
 432         memset(&stream, 0, sizeof(stream));
 433         git_inflate_init(&stream);
 434         stream.next_out = buf;
 435         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 436
 437         do {
 438                 unsigned char *last_out = stream.next_out;
 439                 stream.next_in = fill(1);
 440                 stream.avail_in = input_len;
 441                 status = git_inflate(&stream, 0);
 442                 use(input_len - stream.avail_in);
 443                 if (sha1)
 444                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 445                 if (buf == fixed_buf) {
 446                         stream.next_out = buf;
 447                         stream.avail_out = sizeof(fixed_buf);
 448                 }
 449         } while (status == Z_OK);
 450         if (stream.total_out != size || status != Z_STREAM_END)
 451                 bad_object(offset, _("inflate returned %d"), status);
 452         git_inflate_end(&stream);
 453         if (sha1)
 454                 git_SHA1_Final(sha1, &c);
 455         return buf == fixed_buf ? NULL : buf;
 456 }
 457
 458 static void *unpack_raw_entry(struct object_entry *obj,
 459                               union delta_base *delta_base,
 460                               unsigned char *sha1)
 461 {
 462         unsigned char *p;
 463         unsigned long size, c;
 464         off_t base_offset;
 465         unsigned shift;
 466         void *data;
 467
 468         obj->idx.offset = consumed_bytes;
 469         input_crc32 = crc32(0, NULL, 0);
 470
 471         p = fill(1);
 472         c = *p;
 473         use(1);
 474         obj->type = (c >> 4) & 7;
 475         size = (c & 15);
 476         shift = 4;
 477         while (c & 0x80) {
 478                 p = fill(1);
 479                 c = *p;
 480                 use(1);
 481                 size += (c & 0x7f) << shift;
 482                 shift += 7;
 483         }
 484         obj->size = size;
 485
 486         switch (obj->type) {
 487         case OBJ_REF_DELTA:
 488                 hashcpy(delta_base->sha1, fill(20));
 489                 use(20);
 490                 break;
 491         case OBJ_OFS_DELTA:
 492                 memset(delta_base, 0, sizeof(*delta_base));
 493                 p = fill(1);
 494                 c = *p;
 495                 use(1);
 496                 base_offset = c & 127;
 497                 while (c & 128) {
 498                         base_offset += 1;
 499                         if (!base_offset || MSB(base_offset, 7))
 500                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 501                         p = fill(1);
 502                         c = *p;
 503                         use(1);
 504                         base_offset = (base_offset << 7) + (c & 127);
 505                 }
 506                 delta_base->offset = obj->idx.offset - base_offset;
 507                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 508                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 509                 break;
 510         case OBJ_COMMIT:
 511         case OBJ_TREE:
 512         case OBJ_BLOB:
 513         case OBJ_TAG:
 514                 break;
 515         default:
 516                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 517         }
 518         obj->hdr_size = consumed_bytes - obj->idx.offset;
 519
 520         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 521         obj->idx.crc32 = input_crc32;
 522         return data;
 523 }
 524
 525 static void *unpack_data(struct object_entry *obj,
 526                          int (*consume)(const unsigned char *, unsigned long, void *),
 527                          void *cb_data)
 528 {
 529         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 530         unsigned long len = obj[1].idx.offset - from;
 531         unsigned char *data, *inbuf;
 532         git_zstream stream;
 533         int status;
 534
 535         data = xmalloc(consume ? 64*1024 : obj->size);
 536         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 537
 538         memset(&stream, 0, sizeof(stream));
 539         git_inflate_init(&stream);
 540         stream.next_out = data;
 541         stream.avail_out = consume ? 64*1024 : obj->size;
 542
 543         do {
 544                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 545                 n = pread(pack_fd, inbuf, n, from);
 546                 if (n < 0)
 547                         die_errno(_("cannot pread pack file"));
 548                 if (!n)
 549                         die(Q_("premature end of pack file, %lu byte missing",
 550                                "premature end of pack file, %lu bytes missing",
 551                                len),
 552                             len);
 553                 from += n;
 554                 len -= n;
 555                 stream.next_in = inbuf;
 556                 stream.avail_in = n;
 557                 if (!consume)
 558                         status = git_inflate(&stream, 0);
 559                 else {
 560                         do {
 561                                 status = git_inflate(&stream, 0);
 562                                 if (consume(data, stream.next_out - data, cb_data)) {
 563                                         free(inbuf);
 564                                         free(data);
 565                                         return NULL;
 566                                 }
 567                                 stream.next_out = data;
 568                                 stream.avail_out = 64*1024;
 569                         } while (status == Z_OK && stream.avail_in);
 570                 }
 571         } while (len && status == Z_OK && !stream.avail_in);
 572
 573         /* This has been inflated OK when first encountered, so... */
 574         if (status != Z_STREAM_END || stream.total_out != obj->size)
 575                 die(_("serious inflate inconsistency"));
 576
 577         git_inflate_end(&stream);
 578         free(inbuf);
 579         if (consume) {
 580                 free(data);
 581                 data = NULL;
 582         }
 583         return data;
 584 }
 585
 586 static void *get_data_from_pack(struct object_entry *obj)
 587 {
 588         return unpack_data(obj, NULL, NULL);
 589 }
 590
 591 static int compare_delta_bases(const union delta_base *base1,
 592                                const union delta_base *base2,
 593                                enum object_type type1,
 594                                enum object_type type2)
 595 {
 596         int cmp = type1 - type2;
 597         if (cmp)
 598                 return cmp;
 599         return memcmp(base1, base2, UNION_BASE_SZ);
 600 }
 601
 602 static int find_delta(const union delta_base *base, enum object_type type)
 603 {
 604         int first = 0, last = nr_deltas;
 605
 606         while (first < last) {
 607                 int next = (first + last) / 2;
 608                 struct delta_entry *delta = &deltas[next];
 609                 int cmp;
 610
 611                 cmp = compare_delta_bases(base, &delta->base,
 612                                           type, objects[delta->obj_no].type);
 613                 if (!cmp)
 614                         return next;
 615                 if (cmp < 0) {
 616                         last = next;
 617                         continue;
 618                 }
 619                 first = next+1;
 620         }
 621         return -first-1;
 622 }
 623
 624 static void find_delta_children(const union delta_base *base,
 625                                 int *first_index, int *last_index,
 626                                 enum object_type type)
 627 {
 628         int first = find_delta(base, type);
 629         int last = first;
 630         int end = nr_deltas - 1;
 631
 632         if (first < 0) {
 633                 *first_index = 0;
 634                 *last_index = -1;
 635                 return;
 636         }
 637         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 638                 --first;
 639         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 640                 ++last;
 641         *first_index = first;
 642         *last_index = last;
 643 }
 644
 645 struct compare_data {
 646         struct object_entry *entry;
 647         struct git_istream *st;
 648         unsigned char *buf;
 649         unsigned long buf_size;
 650 };
 651
 652 static int compare_objects(const unsigned char *buf, unsigned long size,
 653                            void *cb_data)
 654 {
 655         struct compare_data *data = cb_data;
 656
 657         if (data->buf_size < size) {
 658                 free(data->buf);
 659                 data->buf = xmalloc(size);
 660                 data->buf_size = size;
 661         }
 662
 663         while (size) {
 664                 ssize_t len = read_istream(data->st, data->buf, size);
 665                 if (len == 0)
 666                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 667                             sha1_to_hex(data->entry->idx.sha1));
 668                 if (len < 0)
 669                         die(_("unable to read %s"),
 670                             sha1_to_hex(data->entry->idx.sha1));
 671                 if (memcmp(buf, data->buf, len))
 672                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 673                             sha1_to_hex(data->entry->idx.sha1));
 674                 size -= len;
 675                 buf += len;
 676         }
 677         return 0;
 678 }
 679
 680 static int check_collison(struct object_entry *entry)
 681 {
 682         struct compare_data data;
 683         enum object_type type;
 684         unsigned long size;
 685
 686         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 687                 return -1;
 688
 689         memset(&data, 0, sizeof(data));
 690         data.entry = entry;
 691         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 692         if (!data.st)
 693                 return -1;
 694         if (size != entry->size || type != entry->type)
 695                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 696                     sha1_to_hex(entry->idx.sha1));
 697         unpack_data(entry, compare_objects, &data);
 698         close_istream(data.st);
 699         free(data.buf);
 700         return 0;
 701 }
 702
 703 static void sha1_object(const void *data, struct object_entry *obj_entry,
 704                         unsigned long size, enum object_type type,
 705                         const unsigned char *sha1)
 706 {
 707         void *new_data = NULL;
 708         int collision_test_needed;
 709
 710         assert(data || obj_entry);
 711
 712         read_lock();
 713         collision_test_needed = has_sha1_file(sha1);
 714         read_unlock();
 715
 716         if (collision_test_needed && !data) {
 717                 read_lock();
 718                 if (!check_collison(obj_entry))
 719                         collision_test_needed = 0;
 720                 read_unlock();
 721         }
 722         if (collision_test_needed) {
 723                 void *has_data;
 724                 enum object_type has_type;
 725                 unsigned long has_size;
 726                 read_lock();
 727                 has_type = sha1_object_info(sha1, &has_size);
 728                 if (has_type != type || has_size != size)
 729                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 730                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 731                 read_unlock();
 732                 if (!data)
 733                         data = new_data = get_data_from_pack(obj_entry);
 734                 if (!has_data)
 735                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 736                 if (size != has_size || type != has_type ||
 737                     memcmp(data, has_data, size) != 0)
 738                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 739                 free(has_data);
 740         }
 741
 742         if (strict) {
 743                 read_lock();
 744                 if (type == OBJ_BLOB) {
 745                         struct blob *blob = lookup_blob(sha1);
 746                         if (blob)
 747                                 blob->object.flags |= FLAG_CHECKED;
 748                         else
 749                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 750                 } else {
 751                         struct object *obj;
 752                         int eaten;
 753                         void *buf = (void *) data;
 754
 755                         assert(data && "data can only be NULL for large _blobs_");
 756
 757                         /*
 758                          * we do not need to free the memory here, as the
 759                          * buf is deleted by the caller.
 760                          */
 761                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 762                         if (!obj)
 763                                 die(_("invalid %s"), typename(type));
 764                         if (do_fsck_object &&
 765                             fsck_object(obj, 1, fsck_error_function))
 766                                 die(_("Error in object"));
 767                         if (fsck_walk(obj, mark_link, NULL))
 768                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 769
 770                         if (obj->type == OBJ_TREE) {
 771                                 struct tree *item = (struct tree *) obj;
 772                                 item->buffer = NULL;
 773                                 obj->parsed = 0;
 774                         }
 775                         if (obj->type == OBJ_COMMIT) {
 776                                 struct commit *commit = (struct commit *) obj;
 777                                 commit->buffer = NULL;
 778                         }
 779                         obj->flags |= FLAG_CHECKED;
 780                 }
 781                 read_unlock();
 782         }
 783
 784         free(new_data);
 785 }
 786
 787 /*
 788  * This function is part of find_unresolved_deltas(). There are two
 789  * walkers going in the opposite ways.
 790  *
 791  * The first one in find_unresolved_deltas() traverses down from
 792  * parent node to children, deflating nodes along the way. However,
 793  * memory for deflated nodes is limited by delta_base_cache_limit, so
 794  * at some point parent node's deflated content may be freed.
 795  *
 796  * The second walker is this function, which goes from current node up
 797  * to top parent if necessary to deflate the node. In normal
 798  * situation, its parent node would be already deflated, so it just
 799  * needs to apply delta.
 800  *
 801  * In the worst case scenario, parent node is no longer deflated because
 802  * we're running out of delta_base_cache_limit; we need to re-deflate
 803  * parents, possibly up to the top base.
 804  *
 805  * All deflated objects here are subject to be freed if we exceed
 806  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 807  * just need to make sure the last node is not freed.
 808  */
 809 static void *get_base_data(struct base_data *c)
 810 {
 811         if (!c->data) {
 812                 struct object_entry *obj = c->obj;
 813                 struct base_data **delta = NULL;
 814                 int delta_nr = 0, delta_alloc = 0;
 815
 816                 while (is_delta_type(c->obj->type) && !c->data) {
 817                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 818                         delta[delta_nr++] = c;
 819                         c = c->base;
 820                 }
 821                 if (!delta_nr) {
 822                         c->data = get_data_from_pack(obj);
 823                         c->size = obj->size;
 824                         get_thread_data()->base_cache_used += c->size;
 825                         prune_base_data(c);
 826                 }
 827                 for (; delta_nr > 0; delta_nr--) {
 828                         void *base, *raw;
 829                         c = delta[delta_nr - 1];
 830                         obj = c->obj;
 831                         base = get_base_data(c->base);
 832                         raw = get_data_from_pack(obj);
 833                         c->data = patch_delta(
 834                                 base, c->base->size,
 835                                 raw, obj->size,
 836                                 &c->size);
 837                         free(raw);
 838                         if (!c->data)
 839                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 840                         get_thread_data()->base_cache_used += c->size;
 841                         prune_base_data(c);
 842                 }
 843                 free(delta);
 844         }
 845         return c->data;
 846 }
 847
 848 static void resolve_delta(struct object_entry *delta_obj,
 849                           struct base_data *base, struct base_data *result)
 850 {
 851         void *base_data, *delta_data;
 852
 853         delta_obj->real_type = base->obj->real_type;
 854         if (show_stat) {
 855                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 856                 deepest_delta_lock();
 857                 if (deepest_delta < delta_obj->delta_depth)
 858                         deepest_delta = delta_obj->delta_depth;
 859                 deepest_delta_unlock();
 860         }
 861         delta_obj->base_object_no = base->obj - objects;
 862         delta_data = get_data_from_pack(delta_obj);
 863         base_data = get_base_data(base);
 864         result->obj = delta_obj;
 865         result->data = patch_delta(base_data, base->size,
 866                                    delta_data, delta_obj->size, &result->size);
 867         free(delta_data);
 868         if (!result->data)
 869                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 870         hash_sha1_file(result->data, result->size,
 871                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 872         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 873                     delta_obj->idx.sha1);
 874         counter_lock();
 875         nr_resolved_deltas++;
 876         counter_unlock();
 877 }
 878
 879 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 880                                                   struct base_data *prev_base)
 881 {
 882         if (base->ref_last == -1 && base->ofs_last == -1) {
 883                 union delta_base base_spec;
 884
 885                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 886                 find_delta_children(&base_spec,
 887                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 888
 889                 memset(&base_spec, 0, sizeof(base_spec));
 890                 base_spec.offset = base->obj->idx.offset;
 891                 find_delta_children(&base_spec,
 892                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 893
 894                 if (base->ref_last == -1 && base->ofs_last == -1) {
 895                         free(base->data);
 896                         return NULL;
 897                 }
 898
 899                 link_base_data(prev_base, base);
 900         }
 901
 902         if (base->ref_first <= base->ref_last) {
 903                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 904                 struct base_data *result = alloc_base_data();
 905
 906                 assert(child->real_type == OBJ_REF_DELTA);
 907                 resolve_delta(child, base, result);
 908                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 909                         free_base_data(base);
 910
 911                 base->ref_first++;
 912                 return result;
 913         }
 914
 915         if (base->ofs_first <= base->ofs_last) {
 916                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 917                 struct base_data *result = alloc_base_data();
 918
 919                 assert(child->real_type == OBJ_OFS_DELTA);
 920                 resolve_delta(child, base, result);
 921                 if (base->ofs_first == base->ofs_last)
 922                         free_base_data(base);
 923
 924                 base->ofs_first++;
 925                 return result;
 926         }
 927
 928         unlink_base_data(base);
 929         return NULL;
 930 }
 931
 932 static void find_unresolved_deltas(struct base_data *base)
 933 {
 934         struct base_data *new_base, *prev_base = NULL;
 935         for (;;) {
 936                 new_base = find_unresolved_deltas_1(base, prev_base);
 937
 938                 if (new_base) {
 939                         prev_base = base;
 940                         base = new_base;
 941                 } else {
 942                         free(base);
 943                         base = prev_base;
 944                         if (!base)
 945                                 return;
 946                         prev_base = base->base;
 947                 }
 948         }
 949 }
 950
 951 static int compare_delta_entry(const void *a, const void *b)
 952 {
 953         const struct delta_entry *delta_a = a;
 954         const struct delta_entry *delta_b = b;
 955
 956         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 957         return compare_delta_bases(&delta_a->base, &delta_b->base,
 958                                    objects[delta_a->obj_no].type,
 959                                    objects[delta_b->obj_no].type);
 960 }
 961
 962 static void resolve_base(struct object_entry *obj)
 963 {
 964         struct base_data *base_obj = alloc_base_data();
 965         base_obj->obj = obj;
 966         base_obj->data = NULL;
 967         find_unresolved_deltas(base_obj);
 968 }
 969
 970 #ifndef NO_PTHREADS
 971 static void *threaded_second_pass(void *data)
 972 {
 973         set_thread_data(data);
 974         for (;;) {
 975                 int i;
 976                 counter_lock();
 977                 display_progress(progress, nr_resolved_deltas);
 978                 counter_unlock();
 979                 work_lock();
 980                 while (nr_dispatched < nr_objects &&
 981                        is_delta_type(objects[nr_dispatched].type))
 982                         nr_dispatched++;
 983                 if (nr_dispatched >= nr_objects) {
 984                         work_unlock();
 985                         break;
 986                 }
 987                 i = nr_dispatched++;
 988                 work_unlock();
 989
 990                 resolve_base(&objects[i]);
 991         }
 992         return NULL;
 993 }
 994 #endif
 995
 996 /*
 997  * First pass:
 998  * - find locations of all objects;
 999  * - calculate SHA1 of all non-delta objects;
1000  * - remember base (SHA1 or offset) for all deltas.
1001  */
1002 static void parse_pack_objects(unsigned char *sha1)
1003 {
1004         int i, nr_delays = 0;
1005         struct delta_entry *delta = deltas;
1006         struct stat st;
1007
1008         if (verbose)
1009                 progress = start_progress(
1010                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1011                                 nr_objects);
1012         for (i = 0; i < nr_objects; i++) {
1013                 struct object_entry *obj = &objects[i];
1014                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1015                 obj->real_type = obj->type;
1016                 if (is_delta_type(obj->type)) {
1017                         nr_deltas++;
1018                         delta->obj_no = i;
1019                         delta++;
1020                 } else if (!data) {
1021                         /* large blobs, check later */
1022                         obj->real_type = OBJ_BAD;
1023                         nr_delays++;
1024                 } else
1025                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1026                 free(data);
1027                 display_progress(progress, i+1);
1028         }
1029         objects[i].idx.offset = consumed_bytes;
1030         stop_progress(&progress);
1031
1032         /* Check pack integrity */
1033         flush();
1034         git_SHA1_Final(sha1, &input_ctx);
1035         if (hashcmp(fill(20), sha1))
1036                 die(_("pack is corrupted (SHA1 mismatch)"));
1037         use(20);
1038
1039         /* If input_fd is a file, we should have reached its end now. */
1040         if (fstat(input_fd, &st))
1041                 die_errno(_("cannot fstat packfile"));
1042         if (S_ISREG(st.st_mode) &&
1043                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1044                 die(_("pack has junk at the end"));
1045
1046         for (i = 0; i < nr_objects; i++) {
1047                 struct object_entry *obj = &objects[i];
1048                 if (obj->real_type != OBJ_BAD)
1049                         continue;
1050                 obj->real_type = obj->type;
1051                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1052                 nr_delays--;
1053         }
1054         if (nr_delays)
1055                 die(_("confusion beyond insanity in parse_pack_objects()"));
1056 }
1057
1058 /*
1059  * Second pass:
1060  * - for all non-delta objects, look if it is used as a base for
1061  *   deltas;
1062  * - if used as a base, uncompress the object and apply all deltas,
1063  *   recursively checking if the resulting object is used as a base
1064  *   for some more deltas.
1065  */
1066 static void resolve_deltas(void)
1067 {
1068         int i;
1069
1070         if (!nr_deltas)
1071                 return;
1072
1073         /* Sort deltas by base SHA1/offset for fast searching */
1074         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1075               compare_delta_entry);
1076
1077         if (verbose)
1078                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1079
1080 #ifndef NO_PTHREADS
1081         nr_dispatched = 0;
1082         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1083                 init_thread();
1084                 for (i = 0; i < nr_threads; i++) {
1085                         int ret = pthread_create(&thread_data[i].thread, NULL,
1086                                                  threaded_second_pass, thread_data + i);
1087                         if (ret)
1088                                 die(_("unable to create thread: %s"),
1089                                     strerror(ret));
1090                 }
1091                 for (i = 0; i < nr_threads; i++)
1092                         pthread_join(thread_data[i].thread, NULL);
1093                 cleanup_thread();
1094                 return;
1095         }
1096 #endif
1097
1098         for (i = 0; i < nr_objects; i++) {
1099                 struct object_entry *obj = &objects[i];
1100
1101                 if (is_delta_type(obj->type))
1102                         continue;
1103                 resolve_base(obj);
1104                 display_progress(progress, nr_resolved_deltas);
1105         }
1106 }
1107
1108 /*
1109  * Third pass:
1110  * - append objects to convert thin pack to full pack if required
1111  * - write the final 20-byte SHA-1
1112  */
1113 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1114 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1115 {
1116         if (nr_deltas == nr_resolved_deltas) {
1117                 stop_progress(&progress);
1118                 /* Flush remaining pack final 20-byte SHA1. */
1119                 flush();
1120                 return;
1121         }
1122
1123         if (fix_thin_pack) {
1124                 struct sha1file *f;
1125                 unsigned char read_sha1[20], tail_sha1[20];
1126                 struct strbuf msg = STRBUF_INIT;
1127                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1128                 int nr_objects_initial = nr_objects;
1129                 if (nr_unresolved <= 0)
1130                         die(_("confusion beyond insanity"));
1131                 objects = xrealloc(objects,
1132                                    (nr_objects + nr_unresolved + 1)
1133                                    * sizeof(*objects));
1134                 memset(objects + nr_objects + 1, 0,
1135                        nr_unresolved * sizeof(*objects));
1136                 f = sha1fd(output_fd, curr_pack);
1137                 fix_unresolved_deltas(f, nr_unresolved);
1138                 strbuf_addf(&msg, _("completed with %d local objects"),
1139                             nr_objects - nr_objects_initial);
1140                 stop_progress_msg(&progress, msg.buf);
1141                 strbuf_release(&msg);
1142                 sha1close(f, tail_sha1, 0);
1143                 hashcpy(read_sha1, pack_sha1);
1144                 fixup_pack_header_footer(output_fd, pack_sha1,
1145                                          curr_pack, nr_objects,
1146                                          read_sha1, consumed_bytes-20);
1147                 if (hashcmp(read_sha1, tail_sha1) != 0)
1148                         die(_("Unexpected tail checksum for %s "
1149                               "(disk corruption?)"), curr_pack);
1150         }
1151         if (nr_deltas != nr_resolved_deltas)
1152                 die(Q_("pack has %d unresolved delta",
1153                        "pack has %d unresolved deltas",
1154                        nr_deltas - nr_resolved_deltas),
1155                     nr_deltas - nr_resolved_deltas);
1156 }
1157
1158 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1159 {
1160         git_zstream stream;
1161         int status;
1162         unsigned char outbuf[4096];
1163
1164         memset(&stream, 0, sizeof(stream));
1165         git_deflate_init(&stream, zlib_compression_level);
1166         stream.next_in = in;
1167         stream.avail_in = size;
1168
1169         do {
1170                 stream.next_out = outbuf;
1171                 stream.avail_out = sizeof(outbuf);
1172                 status = git_deflate(&stream, Z_FINISH);
1173                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1174         } while (status == Z_OK);
1175
1176         if (status != Z_STREAM_END)
1177                 die(_("unable to deflate appended object (%d)"), status);
1178         size = stream.total_out;
1179         git_deflate_end(&stream);
1180         return size;
1181 }
1182
1183 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1184                                const unsigned char *sha1, void *buf,
1185                                unsigned long size, enum object_type type)
1186 {
1187         struct object_entry *obj = &objects[nr_objects++];
1188         unsigned char header[10];
1189         unsigned long s = size;
1190         int n = 0;
1191         unsigned char c = (type << 4) | (s & 15);
1192         s >>= 4;
1193         while (s) {
1194                 header[n++] = c | 0x80;
1195                 c = s & 0x7f;
1196                 s >>= 7;
1197         }
1198         header[n++] = c;
1199         crc32_begin(f);
1200         sha1write(f, header, n);
1201         obj[0].size = size;
1202         obj[0].hdr_size = n;
1203         obj[0].type = type;
1204         obj[0].real_type = type;
1205         obj[1].idx.offset = obj[0].idx.offset + n;
1206         obj[1].idx.offset += write_compressed(f, buf, size);
1207         obj[0].idx.crc32 = crc32_end(f);
1208         sha1flush(f);
1209         hashcpy(obj->idx.sha1, sha1);
1210         return obj;
1211 }
1212
1213 static int delta_pos_compare(const void *_a, const void *_b)
1214 {
1215         struct delta_entry *a = *(struct delta_entry **)_a;
1216         struct delta_entry *b = *(struct delta_entry **)_b;
1217         return a->obj_no - b->obj_no;
1218 }
1219
1220 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1221 {
1222         struct delta_entry **sorted_by_pos;
1223         int i, n = 0;
1224
1225         /*
1226          * Since many unresolved deltas may well be themselves base objects
1227          * for more unresolved deltas, we really want to include the
1228          * smallest number of base objects that would cover as much delta
1229          * as possible by picking the
1230          * trunc deltas first, allowing for other deltas to resolve without
1231          * additional base objects.  Since most base objects are to be found
1232          * before deltas depending on them, a good heuristic is to start
1233          * resolving deltas in the same order as their position in the pack.
1234          */
1235         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1236         for (i = 0; i < nr_deltas; i++) {
1237                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1238                         continue;
1239                 sorted_by_pos[n++] = &deltas[i];
1240         }
1241         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1242
1243         for (i = 0; i < n; i++) {
1244                 struct delta_entry *d = sorted_by_pos[i];
1245                 enum object_type type;
1246                 struct base_data *base_obj = alloc_base_data();
1247
1248                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1249                         continue;
1250                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1251                 if (!base_obj->data)
1252                         continue;
1253
1254                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1255                                 base_obj->size, typename(type)))
1256                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1257                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1258                                         base_obj->data, base_obj->size, type);
1259                 find_unresolved_deltas(base_obj);
1260                 display_progress(progress, nr_resolved_deltas);
1261         }
1262         free(sorted_by_pos);
1263 }
1264
1265 static void final(const char *final_pack_name, const char *curr_pack_name,
1266                   const char *final_index_name, const char *curr_index_name,
1267                   const char *keep_name, const char *keep_msg,
1268                   unsigned char *sha1)
1269 {
1270         const char *report = "pack";
1271         char name[PATH_MAX];
1272         int err;
1273
1274         if (!from_stdin) {
1275                 close(input_fd);
1276         } else {
1277                 fsync_or_die(output_fd, curr_pack_name);
1278                 err = close(output_fd);
1279                 if (err)
1280                         die_errno(_("error while closing pack file"));
1281         }
1282
1283         if (keep_msg) {
1284                 int keep_fd, keep_msg_len = strlen(keep_msg);
1285
1286                 if (!keep_name)
1287                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1288                 else
1289                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1290
1291                 if (keep_fd < 0) {
1292                         if (errno != EEXIST)
1293                                 die_errno(_("cannot write keep file '%s'"),
1294                                           keep_name ? keep_name : name);
1295                 } else {
1296                         if (keep_msg_len > 0) {
1297                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1298                                 write_or_die(keep_fd, "\n", 1);
1299                         }
1300                         if (close(keep_fd) != 0)
1301                                 die_errno(_("cannot close written keep file '%s'"),
1302                                           keep_name ? keep_name : name);
1303                         report = "keep";
1304                 }
1305         }
1306
1307         if (final_pack_name != curr_pack_name) {
1308                 if (!final_pack_name) {
1309                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1310                                  get_object_directory(), sha1_to_hex(sha1));
1311                         final_pack_name = name;
1312                 }
1313                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1314                         die(_("cannot store pack file"));
1315         } else if (from_stdin)
1316                 chmod(final_pack_name, 0444);
1317
1318         if (final_index_name != curr_index_name) {
1319                 if (!final_index_name) {
1320                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1321                                  get_object_directory(), sha1_to_hex(sha1));
1322                         final_index_name = name;
1323                 }
1324                 if (move_temp_to_file(curr_index_name, final_index_name))
1325                         die(_("cannot store index file"));
1326         } else
1327                 chmod(final_index_name, 0444);
1328
1329         if (!from_stdin) {
1330                 printf("%s\n", sha1_to_hex(sha1));
1331         } else {
1332                 char buf[48];
1333                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1334                                    report, sha1_to_hex(sha1));
1335                 write_or_die(1, buf, len);
1336
1337                 /*
1338                  * Let's just mimic git-unpack-objects here and write
1339                  * the last part of the input buffer to stdout.
1340                  */
1341                 while (input_len) {
1342                         err = xwrite(1, input_buffer + input_offset, input_len);
1343                         if (err <= 0)
1344                                 break;
1345                         input_len -= err;
1346                         input_offset += err;
1347                 }
1348         }
1349 }
1350
1351 static int git_index_pack_config(const char *k, const char *v, void *cb)
1352 {
1353         struct pack_idx_option *opts = cb;
1354
1355         if (!strcmp(k, "pack.indexversion")) {
1356                 opts->version = git_config_int(k, v);
1357                 if (opts->version > 2)
1358                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1359                 return 0;
1360         }
1361         if (!strcmp(k, "pack.threads")) {
1362                 nr_threads = git_config_int(k, v);
1363                 if (nr_threads < 0)
1364                         die(_("invalid number of threads specified (%d)"),
1365                             nr_threads);
1366 #ifdef NO_PTHREADS
1367                 if (nr_threads != 1)
1368                         warning(_("no threads support, ignoring %s"), k);
1369                 nr_threads = 1;
1370 #endif
1371                 return 0;
1372         }
1373         return git_default_config(k, v, cb);
1374 }
1375
1376 static int cmp_uint32(const void *a_, const void *b_)
1377 {
1378         uint32_t a = *((uint32_t *)a_);
1379         uint32_t b = *((uint32_t *)b_);
1380
1381         return (a < b) ? -1 : (a != b);
1382 }
1383
1384 static void read_v2_anomalous_offsets(struct packed_git *p,
1385                                       struct pack_idx_option *opts)
1386 {
1387         const uint32_t *idx1, *idx2;
1388         uint32_t i;
1389
1390         /* The address of the 4-byte offset table */
1391         idx1 = (((const uint32_t *)p->index_data)
1392                 + 2 /* 8-byte header */
1393                 + 256 /* fan out */
1394                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1395                 + p->num_objects /* CRC32 table */
1396                 );
1397
1398         /* The address of the 8-byte offset table */
1399         idx2 = idx1 + p->num_objects;
1400
1401         for (i = 0; i < p->num_objects; i++) {
1402                 uint32_t off = ntohl(idx1[i]);
1403                 if (!(off & 0x80000000))
1404                         continue;
1405                 off = off & 0x7fffffff;
1406                 if (idx2[off * 2])
1407                         continue;
1408                 /*
1409                  * The real offset is ntohl(idx2[off * 2]) in high 4
1410                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1411                  * octets.  But idx2[off * 2] is Zero!!!
1412                  */
1413                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1414                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1415         }
1416
1417         if (1 < opts->anomaly_nr)
1418                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1419 }
1420
1421 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1422 {
1423         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1424
1425         if (!p)
1426                 die(_("Cannot open existing pack file '%s'"), pack_name);
1427         if (open_pack_index(p))
1428                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1429
1430         /* Read the attributes from the existing idx file */
1431         opts->version = p->index_version;
1432
1433         if (opts->version == 2)
1434                 read_v2_anomalous_offsets(p, opts);
1435
1436         /*
1437          * Get rid of the idx file as we do not need it anymore.
1438          * NEEDSWORK: extract this bit from free_pack_by_name() in
1439          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1440          * know we haven't installed this pack (hence we never have
1441          * read anything from it).
1442          */
1443         close_pack_index(p);
1444         free(p);
1445 }
1446
1447 static void show_pack_info(int stat_only)
1448 {
1449         int i, baseobjects = nr_objects - nr_deltas;
1450         unsigned long *chain_histogram = NULL;
1451
1452         if (deepest_delta)
1453                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1454
1455         for (i = 0; i < nr_objects; i++) {
1456                 struct object_entry *obj = &objects[i];
1457
1458                 if (is_delta_type(obj->type))
1459                         chain_histogram[obj->delta_depth - 1]++;
1460                 if (stat_only)
1461                         continue;
1462                 printf("%s %-6s %lu %lu %"PRIuMAX,
1463                        sha1_to_hex(obj->idx.sha1),
1464                        typename(obj->real_type), obj->size,
1465                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1466                        (uintmax_t)obj->idx.offset);
1467                 if (is_delta_type(obj->type)) {
1468                         struct object_entry *bobj = &objects[obj->base_object_no];
1469                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1470                 }
1471                 putchar('\n');
1472         }
1473
1474         if (baseobjects)
1475                 printf_ln(Q_("non delta: %d object",
1476                              "non delta: %d objects",
1477                              baseobjects),
1478                           baseobjects);
1479         for (i = 0; i < deepest_delta; i++) {
1480                 if (!chain_histogram[i])
1481                         continue;
1482                 printf_ln(Q_("chain length = %d: %lu object",
1483                              "chain length = %d: %lu objects",
1484                              chain_histogram[i]),
1485                           i + 1,
1486                           chain_histogram[i]);
1487         }
1488 }
1489
1490 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1491 {
1492         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1493         const char *curr_pack, *curr_index;
1494         const char *index_name = NULL, *pack_name = NULL;
1495         const char *keep_name = NULL, *keep_msg = NULL;
1496         char *index_name_buf = NULL, *keep_name_buf = NULL;
1497         struct pack_idx_entry **idx_objects;
1498         struct pack_idx_option opts;
1499         unsigned char pack_sha1[20];
1500         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1501
1502         if (argc == 2 && !strcmp(argv[1], "-h"))
1503                 usage(index_pack_usage);
1504
1505         read_replace_refs = 0;
1506
1507         reset_pack_idx_option(&opts);
1508         git_config(git_index_pack_config, &opts);
1509         if (prefix && chdir(prefix))
1510                 die(_("Cannot come back to cwd"));
1511
1512         for (i = 1; i < argc; i++) {
1513                 const char *arg = argv[i];
1514
1515                 if (*arg == '-') {
1516                         if (!strcmp(arg, "--stdin")) {
1517                                 from_stdin = 1;
1518                         } else if (!strcmp(arg, "--fix-thin")) {
1519                                 fix_thin_pack = 1;
1520                         } else if (!strcmp(arg, "--strict")) {
1521                                 strict = 1;
1522                                 do_fsck_object = 1;
1523                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1524                                 strict = 1;
1525                                 check_self_contained_and_connected = 1;
1526                         } else if (!strcmp(arg, "--verify")) {
1527                                 verify = 1;
1528                         } else if (!strcmp(arg, "--verify-stat")) {
1529                                 verify = 1;
1530                                 show_stat = 1;
1531                         } else if (!strcmp(arg, "--verify-stat-only")) {
1532                                 verify = 1;
1533                                 show_stat = 1;
1534                                 stat_only = 1;
1535                         } else if (!strcmp(arg, "--keep")) {
1536                                 keep_msg = "";
1537                         } else if (starts_with(arg, "--keep=")) {
1538                                 keep_msg = arg + 7;
1539                         } else if (starts_with(arg, "--threads=")) {
1540                                 char *end;
1541                                 nr_threads = strtoul(arg+10, &end, 0);
1542                                 if (!arg[10] || *end || nr_threads < 0)
1543                                         usage(index_pack_usage);
1544 #ifdef NO_PTHREADS
1545                                 if (nr_threads != 1)
1546                                         warning(_("no threads support, "
1547                                                   "ignoring %s"), arg);
1548                                 nr_threads = 1;
1549 #endif
1550                         } else if (starts_with(arg, "--pack_header=")) {
1551                                 struct pack_header *hdr;
1552                                 char *c;
1553
1554                                 hdr = (struct pack_header *)input_buffer;
1555                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1556                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1557                                 if (*c != ',')
1558                                         die(_("bad %s"), arg);
1559                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1560                                 if (*c)
1561                                         die(_("bad %s"), arg);
1562                                 input_len = sizeof(*hdr);
1563                         } else if (!strcmp(arg, "-v")) {
1564                                 verbose = 1;
1565                         } else if (!strcmp(arg, "-o")) {
1566                                 if (index_name || (i+1) >= argc)
1567                                         usage(index_pack_usage);
1568                                 index_name = argv[++i];
1569                         } else if (starts_with(arg, "--index-version=")) {
1570                                 char *c;
1571                                 opts.version = strtoul(arg + 16, &c, 10);
1572                                 if (opts.version > 2)
1573                                         die(_("bad %s"), arg);
1574                                 if (*c == ',')
1575                                         opts.off32_limit = strtoul(c+1, &c, 0);
1576                                 if (*c || opts.off32_limit & 0x80000000)
1577                                         die(_("bad %s"), arg);
1578                         } else
1579                                 usage(index_pack_usage);
1580                         continue;
1581                 }
1582
1583                 if (pack_name)
1584                         usage(index_pack_usage);
1585                 pack_name = arg;
1586         }
1587
1588         if (!pack_name && !from_stdin)
1589                 usage(index_pack_usage);
1590         if (fix_thin_pack && !from_stdin)
1591                 die(_("--fix-thin cannot be used without --stdin"));
1592         if (!index_name && pack_name) {
1593                 int len = strlen(pack_name);
1594                 if (!has_extension(pack_name, ".pack"))
1595                         die(_("packfile name '%s' does not end with '.pack'"),
1596                             pack_name);
1597                 index_name_buf = xmalloc(len);
1598                 memcpy(index_name_buf, pack_name, len - 5);
1599                 strcpy(index_name_buf + len - 5, ".idx");
1600                 index_name = index_name_buf;
1601         }
1602         if (keep_msg && !keep_name && pack_name) {
1603                 int len = strlen(pack_name);
1604                 if (!has_extension(pack_name, ".pack"))
1605                         die(_("packfile name '%s' does not end with '.pack'"),
1606                             pack_name);
1607                 keep_name_buf = xmalloc(len);
1608                 memcpy(keep_name_buf, pack_name, len - 5);
1609                 strcpy(keep_name_buf + len - 5, ".keep");
1610                 keep_name = keep_name_buf;
1611         }
1612         if (verify) {
1613                 if (!index_name)
1614                         die(_("--verify with no packfile name given"));
1615                 read_idx_option(&opts, index_name);
1616                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1617         }
1618         if (strict)
1619                 opts.flags |= WRITE_IDX_STRICT;
1620
1621 #ifndef NO_PTHREADS
1622         if (!nr_threads) {
1623                 nr_threads = online_cpus();
1624                 /* An experiment showed that more threads does not mean faster */
1625                 if (nr_threads > 3)
1626                         nr_threads = 3;
1627         }
1628 #endif
1629
1630         curr_pack = open_pack_file(pack_name);
1631         parse_pack_header();
1632         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1633         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1634         parse_pack_objects(pack_sha1);
1635         resolve_deltas();
1636         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1637         free(deltas);
1638         if (strict)
1639                 foreign_nr = check_objects();
1640
1641         if (show_stat)
1642                 show_pack_info(stat_only);
1643
1644         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1645         for (i = 0; i < nr_objects; i++)
1646                 idx_objects[i] = &objects[i].idx;
1647         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1648         free(idx_objects);
1649
1650         if (!verify)
1651                 final(pack_name, curr_pack,
1652                       index_name, curr_index,
1653                       keep_name, keep_msg,
1654                       pack_sha1);
1655         else
1656                 close(input_fd);
1657         free(objects);
1658         free(index_name_buf);
1659         free(keep_name_buf);
1660         if (pack_name == NULL)
1661                 free((void *) curr_pack);
1662         if (index_name == NULL)
1663                 free((void *) curr_index);
1664
1665         /*
1666          * Let the caller know this pack is not self contained
1667          */
1668         if (check_self_contained_and_connected && foreign_nr)
1669                 return 1;
1670
1671         return 0;
1672 }