git.oblomov.eu Git - git/blob - builtin/index-pack.c

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned int hdr_size;
  22         enum object_type type;
  23         enum object_type real_type;
  24         unsigned delta_depth;
  25         int base_object_no;
  26 };
  27
  28 union delta_base {
  29         unsigned char sha1[20];
  30         off_t offset;
  31 };
  32
  33 struct base_data {
  34         struct base_data *base;
  35         struct base_data *child;
  36         struct object_entry *obj;
  37         void *data;
  38         unsigned long size;
  39         int ref_first, ref_last;
  40         int ofs_first, ofs_last;
  41 };
  42
  43 struct thread_local {
  44 #ifndef NO_PTHREADS
  45         pthread_t thread;
  46 #endif
  47         struct base_data *base_cache;
  48         size_t base_cache_used;
  49         int pack_fd;
  50 };
  51
  52 /*
  53  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  54  * to memcmp() only the first 20 bytes.
  55  */
  56 #define UNION_BASE_SZ   20
  57
  58 #define FLAG_LINK (1u<<20)
  59 #define FLAG_CHECKED (1u<<21)
  60
  61 struct delta_entry {
  62         union delta_base base;
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct delta_entry *deltas;
  68 static struct thread_local nothread_data;
  69 static int nr_objects;
  70 static int nr_deltas;
  71 static int nr_resolved_deltas;
  72 static int nr_threads;
  73
  74 static int from_stdin;
  75 static int strict;
  76 static int do_fsck_object;
  77 static int verbose;
  78 static int show_stat;
  79 static int check_self_contained_and_connected;
  80
  81 static struct progress *progress;
  82
  83 /* We always read in 4kB chunks. */
  84 static unsigned char input_buffer[4096];
  85 static unsigned int input_offset, input_len;
  86 static off_t consumed_bytes;
  87 static unsigned deepest_delta;
  88 static git_SHA_CTX input_ctx;
  89 static uint32_t input_crc32;
  90 static int input_fd, output_fd;
  91 static const char *curr_pack;
  92
  93 #ifndef NO_PTHREADS
  94
  95 static struct thread_local *thread_data;
  96 static int nr_dispatched;
  97 static int threads_active;
  98
  99 static pthread_mutex_t read_mutex;
 100 #define read_lock()             lock_mutex(&read_mutex)
 101 #define read_unlock()           unlock_mutex(&read_mutex)
 102
 103 static pthread_mutex_t counter_mutex;
 104 #define counter_lock()          lock_mutex(&counter_mutex)
 105 #define counter_unlock()        unlock_mutex(&counter_mutex)
 106
 107 static pthread_mutex_t work_mutex;
 108 #define work_lock()             lock_mutex(&work_mutex)
 109 #define work_unlock()           unlock_mutex(&work_mutex)
 110
 111 static pthread_mutex_t deepest_delta_mutex;
 112 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 113 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 114
 115 static pthread_key_t key;
 116
 117 static inline void lock_mutex(pthread_mutex_t *mutex)
 118 {
 119         if (threads_active)
 120                 pthread_mutex_lock(mutex);
 121 }
 122
 123 static inline void unlock_mutex(pthread_mutex_t *mutex)
 124 {
 125         if (threads_active)
 126                 pthread_mutex_unlock(mutex);
 127 }
 128
 129 /*
 130  * Mutex and conditional variable can't be statically-initialized on Windows.
 131  */
 132 static void init_thread(void)
 133 {
 134         int i;
 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         for (i = 0; i < nr_threads; i++) {
 143                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 144                 if (thread_data[i].pack_fd == -1)
 145                         die_errno(_("unable to open %s"), curr_pack);
 146         }
 147
 148         threads_active = 1;
 149 }
 150
 151 static void cleanup_thread(void)
 152 {
 153         int i;
 154         if (!threads_active)
 155                 return;
 156         threads_active = 0;
 157         pthread_mutex_destroy(&read_mutex);
 158         pthread_mutex_destroy(&counter_mutex);
 159         pthread_mutex_destroy(&work_mutex);
 160         if (show_stat)
 161                 pthread_mutex_destroy(&deepest_delta_mutex);
 162         for (i = 0; i < nr_threads; i++)
 163                 close(thread_data[i].pack_fd);
 164         pthread_key_delete(key);
 165         free(thread_data);
 166 }
 167
 168 #else
 169
 170 #define read_lock()
 171 #define read_unlock()
 172
 173 #define counter_lock()
 174 #define counter_unlock()
 175
 176 #define work_lock()
 177 #define work_unlock()
 178
 179 #define deepest_delta_lock()
 180 #define deepest_delta_unlock()
 181
 182 #endif
 183
 184
 185 static int mark_link(struct object *obj, int type, void *data)
 186 {
 187         if (!obj)
 188                 return -1;
 189
 190         if (type != OBJ_ANY && obj->type != type)
 191                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 192
 193         obj->flags |= FLAG_LINK;
 194         return 0;
 195 }
 196
 197 /* The content of each linked object must have been checked
 198    or it must be already present in the object database */
 199 static unsigned check_object(struct object *obj)
 200 {
 201         if (!obj)
 202                 return 0;
 203
 204         if (!(obj->flags & FLAG_LINK))
 205                 return 0;
 206
 207         if (!(obj->flags & FLAG_CHECKED)) {
 208                 unsigned long size;
 209                 int type = sha1_object_info(obj->sha1, &size);
 210                 if (type <= 0)
 211                         die(_("did not receive expected object %s"),
 212                               sha1_to_hex(obj->sha1));
 213                 if (type != obj->type)
 214                         die(_("object %s: expected type %s, found %s"),
 215                             sha1_to_hex(obj->sha1),
 216                             typename(obj->type), typename(type));
 217                 obj->flags |= FLAG_CHECKED;
 218                 return 1;
 219         }
 220
 221         return 0;
 222 }
 223
 224 static unsigned check_objects(void)
 225 {
 226         unsigned i, max, foreign_nr = 0;
 227
 228         max = get_max_object_index();
 229         for (i = 0; i < max; i++)
 230                 foreign_nr += check_object(get_indexed_object(i));
 231         return foreign_nr;
 232 }
 233
 234
 235 /* Discard current buffer used content. */
 236 static void flush(void)
 237 {
 238         if (input_offset) {
 239                 if (output_fd >= 0)
 240                         write_or_die(output_fd, input_buffer, input_offset);
 241                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 242                 memmove(input_buffer, input_buffer + input_offset, input_len);
 243                 input_offset = 0;
 244         }
 245 }
 246
 247 /*
 248  * Make sure at least "min" bytes are available in the buffer, and
 249  * return the pointer to the buffer.
 250  */
 251 static void *fill(int min)
 252 {
 253         if (min <= input_len)
 254                 return input_buffer + input_offset;
 255         if (min > sizeof(input_buffer))
 256                 die(Q_("cannot fill %d byte",
 257                        "cannot fill %d bytes",
 258                        min),
 259                     min);
 260         flush();
 261         do {
 262                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 263                                 sizeof(input_buffer) - input_len);
 264                 if (ret <= 0) {
 265                         if (!ret)
 266                                 die(_("early EOF"));
 267                         die_errno(_("read error on input"));
 268                 }
 269                 input_len += ret;
 270                 if (from_stdin)
 271                         display_throughput(progress, consumed_bytes + input_len);
 272         } while (input_len < min);
 273         return input_buffer;
 274 }
 275
 276 static void use(int bytes)
 277 {
 278         if (bytes > input_len)
 279                 die(_("used more bytes than were available"));
 280         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 281         input_len -= bytes;
 282         input_offset += bytes;
 283
 284         /* make sure off_t is sufficiently large not to wrap */
 285         if (signed_add_overflows(consumed_bytes, bytes))
 286                 die(_("pack too large for current definition of off_t"));
 287         consumed_bytes += bytes;
 288 }
 289
 290 static const char *open_pack_file(const char *pack_name)
 291 {
 292         if (from_stdin) {
 293                 input_fd = 0;
 294                 if (!pack_name) {
 295                         static char tmp_file[PATH_MAX];
 296                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 297                                                 "pack/tmp_pack_XXXXXX");
 298                         pack_name = xstrdup(tmp_file);
 299                 } else
 300                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 301                 if (output_fd < 0)
 302                         die_errno(_("unable to create '%s'"), pack_name);
 303                 nothread_data.pack_fd = output_fd;
 304         } else {
 305                 input_fd = open(pack_name, O_RDONLY);
 306                 if (input_fd < 0)
 307                         die_errno(_("cannot open packfile '%s'"), pack_name);
 308                 output_fd = -1;
 309                 nothread_data.pack_fd = input_fd;
 310         }
 311         git_SHA1_Init(&input_ctx);
 312         return pack_name;
 313 }
 314
 315 static void parse_pack_header(void)
 316 {
 317         struct pack_header *hdr = fill(sizeof(struct pack_header));
 318
 319         /* Header consistency check */
 320         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 321                 die(_("pack signature mismatch"));
 322         if (!pack_version_ok(hdr->hdr_version))
 323                 die(_("pack version %"PRIu32" unsupported"),
 324                         ntohl(hdr->hdr_version));
 325
 326         nr_objects = ntohl(hdr->hdr_entries);
 327         use(sizeof(struct pack_header));
 328 }
 329
 330 static NORETURN void bad_object(unsigned long offset, const char *format,
 331                        ...) __attribute__((format (printf, 2, 3)));
 332
 333 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 334 {
 335         va_list params;
 336         char buf[1024];
 337
 338         va_start(params, format);
 339         vsnprintf(buf, sizeof(buf), format, params);
 340         va_end(params);
 341         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 342 }
 343
 344 static inline struct thread_local *get_thread_data(void)
 345 {
 346 #ifndef NO_PTHREADS
 347         if (threads_active)
 348                 return pthread_getspecific(key);
 349         assert(!threads_active &&
 350                "This should only be reached when all threads are gone");
 351 #endif
 352         return &nothread_data;
 353 }
 354
 355 #ifndef NO_PTHREADS
 356 static void set_thread_data(struct thread_local *data)
 357 {
 358         if (threads_active)
 359                 pthread_setspecific(key, data);
 360 }
 361 #endif
 362
 363 static struct base_data *alloc_base_data(void)
 364 {
 365         struct base_data *base = xmalloc(sizeof(struct base_data));
 366         memset(base, 0, sizeof(*base));
 367         base->ref_last = -1;
 368         base->ofs_last = -1;
 369         return base;
 370 }
 371
 372 static void free_base_data(struct base_data *c)
 373 {
 374         if (c->data) {
 375                 free(c->data);
 376                 c->data = NULL;
 377                 get_thread_data()->base_cache_used -= c->size;
 378         }
 379 }
 380
 381 static void prune_base_data(struct base_data *retain)
 382 {
 383         struct base_data *b;
 384         struct thread_local *data = get_thread_data();
 385         for (b = data->base_cache;
 386              data->base_cache_used > delta_base_cache_limit && b;
 387              b = b->child) {
 388                 if (b->data && b != retain)
 389                         free_base_data(b);
 390         }
 391 }
 392
 393 static void link_base_data(struct base_data *base, struct base_data *c)
 394 {
 395         if (base)
 396                 base->child = c;
 397         else
 398                 get_thread_data()->base_cache = c;
 399
 400         c->base = base;
 401         c->child = NULL;
 402         if (c->data)
 403                 get_thread_data()->base_cache_used += c->size;
 404         prune_base_data(c);
 405 }
 406
 407 static void unlink_base_data(struct base_data *c)
 408 {
 409         struct base_data *base = c->base;
 410         if (base)
 411                 base->child = NULL;
 412         else
 413                 get_thread_data()->base_cache = NULL;
 414         free_base_data(c);
 415 }
 416
 417 static int is_delta_type(enum object_type type)
 418 {
 419         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 420 }
 421
 422 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 423                                enum object_type type, unsigned char *sha1)
 424 {
 425         static char fixed_buf[8192];
 426         int status;
 427         git_zstream stream;
 428         void *buf;
 429         git_SHA_CTX c;
 430         char hdr[32];
 431         int hdrlen;
 432
 433         if (!is_delta_type(type)) {
 434                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 435                 git_SHA1_Init(&c);
 436                 git_SHA1_Update(&c, hdr, hdrlen);
 437         } else
 438                 sha1 = NULL;
 439         if (type == OBJ_BLOB && size > big_file_threshold)
 440                 buf = fixed_buf;
 441         else
 442                 buf = xmalloc(size);
 443
 444         memset(&stream, 0, sizeof(stream));
 445         git_inflate_init(&stream);
 446         stream.next_out = buf;
 447         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 448
 449         do {
 450                 unsigned char *last_out = stream.next_out;
 451                 stream.next_in = fill(1);
 452                 stream.avail_in = input_len;
 453                 status = git_inflate(&stream, 0);
 454                 use(input_len - stream.avail_in);
 455                 if (sha1)
 456                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 457                 if (buf == fixed_buf) {
 458                         stream.next_out = buf;
 459                         stream.avail_out = sizeof(fixed_buf);
 460                 }
 461         } while (status == Z_OK);
 462         if (stream.total_out != size || status != Z_STREAM_END)
 463                 bad_object(offset, _("inflate returned %d"), status);
 464         git_inflate_end(&stream);
 465         if (sha1)
 466                 git_SHA1_Final(sha1, &c);
 467         return buf == fixed_buf ? NULL : buf;
 468 }
 469
 470 static void *unpack_raw_entry(struct object_entry *obj,
 471                               union delta_base *delta_base,
 472                               unsigned char *sha1)
 473 {
 474         unsigned char *p;
 475         unsigned long size, c;
 476         off_t base_offset;
 477         unsigned shift;
 478         void *data;
 479
 480         obj->idx.offset = consumed_bytes;
 481         input_crc32 = crc32(0, NULL, 0);
 482
 483         p = fill(1);
 484         c = *p;
 485         use(1);
 486         obj->type = (c >> 4) & 7;
 487         size = (c & 15);
 488         shift = 4;
 489         while (c & 0x80) {
 490                 p = fill(1);
 491                 c = *p;
 492                 use(1);
 493                 size += (c & 0x7f) << shift;
 494                 shift += 7;
 495         }
 496         obj->size = size;
 497
 498         switch (obj->type) {
 499         case OBJ_REF_DELTA:
 500                 hashcpy(delta_base->sha1, fill(20));
 501                 use(20);
 502                 break;
 503         case OBJ_OFS_DELTA:
 504                 memset(delta_base, 0, sizeof(*delta_base));
 505                 p = fill(1);
 506                 c = *p;
 507                 use(1);
 508                 base_offset = c & 127;
 509                 while (c & 128) {
 510                         base_offset += 1;
 511                         if (!base_offset || MSB(base_offset, 7))
 512                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 513                         p = fill(1);
 514                         c = *p;
 515                         use(1);
 516                         base_offset = (base_offset << 7) + (c & 127);
 517                 }
 518                 delta_base->offset = obj->idx.offset - base_offset;
 519                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 520                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 521                 break;
 522         case OBJ_COMMIT:
 523         case OBJ_TREE:
 524         case OBJ_BLOB:
 525         case OBJ_TAG:
 526                 break;
 527         default:
 528                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 529         }
 530         obj->hdr_size = consumed_bytes - obj->idx.offset;
 531
 532         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 533         obj->idx.crc32 = input_crc32;
 534         return data;
 535 }
 536
 537 static void *unpack_data(struct object_entry *obj,
 538                          int (*consume)(const unsigned char *, unsigned long, void *),
 539                          void *cb_data)
 540 {
 541         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 542         unsigned long len = obj[1].idx.offset - from;
 543         unsigned char *data, *inbuf;
 544         git_zstream stream;
 545         int status;
 546
 547         data = xmalloc(consume ? 64*1024 : obj->size);
 548         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 549
 550         memset(&stream, 0, sizeof(stream));
 551         git_inflate_init(&stream);
 552         stream.next_out = data;
 553         stream.avail_out = consume ? 64*1024 : obj->size;
 554
 555         do {
 556                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 557                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 558                 if (n < 0)
 559                         die_errno(_("cannot pread pack file"));
 560                 if (!n)
 561                         die(Q_("premature end of pack file, %lu byte missing",
 562                                "premature end of pack file, %lu bytes missing",
 563                                len),
 564                             len);
 565                 from += n;
 566                 len -= n;
 567                 stream.next_in = inbuf;
 568                 stream.avail_in = n;
 569                 if (!consume)
 570                         status = git_inflate(&stream, 0);
 571                 else {
 572                         do {
 573                                 status = git_inflate(&stream, 0);
 574                                 if (consume(data, stream.next_out - data, cb_data)) {
 575                                         free(inbuf);
 576                                         free(data);
 577                                         return NULL;
 578                                 }
 579                                 stream.next_out = data;
 580                                 stream.avail_out = 64*1024;
 581                         } while (status == Z_OK && stream.avail_in);
 582                 }
 583         } while (len && status == Z_OK && !stream.avail_in);
 584
 585         /* This has been inflated OK when first encountered, so... */
 586         if (status != Z_STREAM_END || stream.total_out != obj->size)
 587                 die(_("serious inflate inconsistency"));
 588
 589         git_inflate_end(&stream);
 590         free(inbuf);
 591         if (consume) {
 592                 free(data);
 593                 data = NULL;
 594         }
 595         return data;
 596 }
 597
 598 static void *get_data_from_pack(struct object_entry *obj)
 599 {
 600         return unpack_data(obj, NULL, NULL);
 601 }
 602
 603 static int compare_delta_bases(const union delta_base *base1,
 604                                const union delta_base *base2,
 605                                enum object_type type1,
 606                                enum object_type type2)
 607 {
 608         int cmp = type1 - type2;
 609         if (cmp)
 610                 return cmp;
 611         return memcmp(base1, base2, UNION_BASE_SZ);
 612 }
 613
 614 static int find_delta(const union delta_base *base, enum object_type type)
 615 {
 616         int first = 0, last = nr_deltas;
 617
 618         while (first < last) {
 619                 int next = (first + last) / 2;
 620                 struct delta_entry *delta = &deltas[next];
 621                 int cmp;
 622
 623                 cmp = compare_delta_bases(base, &delta->base,
 624                                           type, objects[delta->obj_no].type);
 625                 if (!cmp)
 626                         return next;
 627                 if (cmp < 0) {
 628                         last = next;
 629                         continue;
 630                 }
 631                 first = next+1;
 632         }
 633         return -first-1;
 634 }
 635
 636 static void find_delta_children(const union delta_base *base,
 637                                 int *first_index, int *last_index,
 638                                 enum object_type type)
 639 {
 640         int first = find_delta(base, type);
 641         int last = first;
 642         int end = nr_deltas - 1;
 643
 644         if (first < 0) {
 645                 *first_index = 0;
 646                 *last_index = -1;
 647                 return;
 648         }
 649         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 650                 --first;
 651         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 652                 ++last;
 653         *first_index = first;
 654         *last_index = last;
 655 }
 656
 657 struct compare_data {
 658         struct object_entry *entry;
 659         struct git_istream *st;
 660         unsigned char *buf;
 661         unsigned long buf_size;
 662 };
 663
 664 static int compare_objects(const unsigned char *buf, unsigned long size,
 665                            void *cb_data)
 666 {
 667         struct compare_data *data = cb_data;
 668
 669         if (data->buf_size < size) {
 670                 free(data->buf);
 671                 data->buf = xmalloc(size);
 672                 data->buf_size = size;
 673         }
 674
 675         while (size) {
 676                 ssize_t len = read_istream(data->st, data->buf, size);
 677                 if (len == 0)
 678                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 679                             sha1_to_hex(data->entry->idx.sha1));
 680                 if (len < 0)
 681                         die(_("unable to read %s"),
 682                             sha1_to_hex(data->entry->idx.sha1));
 683                 if (memcmp(buf, data->buf, len))
 684                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 685                             sha1_to_hex(data->entry->idx.sha1));
 686                 size -= len;
 687                 buf += len;
 688         }
 689         return 0;
 690 }
 691
 692 static int check_collison(struct object_entry *entry)
 693 {
 694         struct compare_data data;
 695         enum object_type type;
 696         unsigned long size;
 697
 698         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 699                 return -1;
 700
 701         memset(&data, 0, sizeof(data));
 702         data.entry = entry;
 703         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 704         if (!data.st)
 705                 return -1;
 706         if (size != entry->size || type != entry->type)
 707                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 708                     sha1_to_hex(entry->idx.sha1));
 709         unpack_data(entry, compare_objects, &data);
 710         close_istream(data.st);
 711         free(data.buf);
 712         return 0;
 713 }
 714
 715 static void sha1_object(const void *data, struct object_entry *obj_entry,
 716                         unsigned long size, enum object_type type,
 717                         const unsigned char *sha1)
 718 {
 719         void *new_data = NULL;
 720         int collision_test_needed;
 721
 722         assert(data || obj_entry);
 723
 724         read_lock();
 725         collision_test_needed = has_sha1_file(sha1);
 726         read_unlock();
 727
 728         if (collision_test_needed && !data) {
 729                 read_lock();
 730                 if (!check_collison(obj_entry))
 731                         collision_test_needed = 0;
 732                 read_unlock();
 733         }
 734         if (collision_test_needed) {
 735                 void *has_data;
 736                 enum object_type has_type;
 737                 unsigned long has_size;
 738                 read_lock();
 739                 has_type = sha1_object_info(sha1, &has_size);
 740                 if (has_type != type || has_size != size)
 741                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 742                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 743                 read_unlock();
 744                 if (!data)
 745                         data = new_data = get_data_from_pack(obj_entry);
 746                 if (!has_data)
 747                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 748                 if (size != has_size || type != has_type ||
 749                     memcmp(data, has_data, size) != 0)
 750                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 751                 free(has_data);
 752         }
 753
 754         if (strict) {
 755                 read_lock();
 756                 if (type == OBJ_BLOB) {
 757                         struct blob *blob = lookup_blob(sha1);
 758                         if (blob)
 759                                 blob->object.flags |= FLAG_CHECKED;
 760                         else
 761                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 762                 } else {
 763                         struct object *obj;
 764                         int eaten;
 765                         void *buf = (void *) data;
 766
 767                         assert(data && "data can only be NULL for large _blobs_");
 768
 769                         /*
 770                          * we do not need to free the memory here, as the
 771                          * buf is deleted by the caller.
 772                          */
 773                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 774                         if (!obj)
 775                                 die(_("invalid %s"), typename(type));
 776                         if (do_fsck_object &&
 777                             fsck_object(obj, 1, fsck_error_function))
 778                                 die(_("Error in object"));
 779                         if (fsck_walk(obj, mark_link, NULL))
 780                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 781
 782                         if (obj->type == OBJ_TREE) {
 783                                 struct tree *item = (struct tree *) obj;
 784                                 item->buffer = NULL;
 785                                 obj->parsed = 0;
 786                         }
 787                         if (obj->type == OBJ_COMMIT) {
 788                                 struct commit *commit = (struct commit *) obj;
 789                                 if (detach_commit_buffer(commit, NULL) != data)
 790                                         die("BUG: parse_object_buffer transmogrified our buffer");
 791                         }
 792                         obj->flags |= FLAG_CHECKED;
 793                 }
 794                 read_unlock();
 795         }
 796
 797         free(new_data);
 798 }
 799
 800 /*
 801  * This function is part of find_unresolved_deltas(). There are two
 802  * walkers going in the opposite ways.
 803  *
 804  * The first one in find_unresolved_deltas() traverses down from
 805  * parent node to children, deflating nodes along the way. However,
 806  * memory for deflated nodes is limited by delta_base_cache_limit, so
 807  * at some point parent node's deflated content may be freed.
 808  *
 809  * The second walker is this function, which goes from current node up
 810  * to top parent if necessary to deflate the node. In normal
 811  * situation, its parent node would be already deflated, so it just
 812  * needs to apply delta.
 813  *
 814  * In the worst case scenario, parent node is no longer deflated because
 815  * we're running out of delta_base_cache_limit; we need to re-deflate
 816  * parents, possibly up to the top base.
 817  *
 818  * All deflated objects here are subject to be freed if we exceed
 819  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 820  * just need to make sure the last node is not freed.
 821  */
 822 static void *get_base_data(struct base_data *c)
 823 {
 824         if (!c->data) {
 825                 struct object_entry *obj = c->obj;
 826                 struct base_data **delta = NULL;
 827                 int delta_nr = 0, delta_alloc = 0;
 828
 829                 while (is_delta_type(c->obj->type) && !c->data) {
 830                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 831                         delta[delta_nr++] = c;
 832                         c = c->base;
 833                 }
 834                 if (!delta_nr) {
 835                         c->data = get_data_from_pack(obj);
 836                         c->size = obj->size;
 837                         get_thread_data()->base_cache_used += c->size;
 838                         prune_base_data(c);
 839                 }
 840                 for (; delta_nr > 0; delta_nr--) {
 841                         void *base, *raw;
 842                         c = delta[delta_nr - 1];
 843                         obj = c->obj;
 844                         base = get_base_data(c->base);
 845                         raw = get_data_from_pack(obj);
 846                         c->data = patch_delta(
 847                                 base, c->base->size,
 848                                 raw, obj->size,
 849                                 &c->size);
 850                         free(raw);
 851                         if (!c->data)
 852                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 853                         get_thread_data()->base_cache_used += c->size;
 854                         prune_base_data(c);
 855                 }
 856                 free(delta);
 857         }
 858         return c->data;
 859 }
 860
 861 static void resolve_delta(struct object_entry *delta_obj,
 862                           struct base_data *base, struct base_data *result)
 863 {
 864         void *base_data, *delta_data;
 865
 866         delta_obj->real_type = base->obj->real_type;
 867         if (show_stat) {
 868                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 869                 deepest_delta_lock();
 870                 if (deepest_delta < delta_obj->delta_depth)
 871                         deepest_delta = delta_obj->delta_depth;
 872                 deepest_delta_unlock();
 873         }
 874         delta_obj->base_object_no = base->obj - objects;
 875         delta_data = get_data_from_pack(delta_obj);
 876         base_data = get_base_data(base);
 877         result->obj = delta_obj;
 878         result->data = patch_delta(base_data, base->size,
 879                                    delta_data, delta_obj->size, &result->size);
 880         free(delta_data);
 881         if (!result->data)
 882                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 883         hash_sha1_file(result->data, result->size,
 884                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 885         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 886                     delta_obj->idx.sha1);
 887         counter_lock();
 888         nr_resolved_deltas++;
 889         counter_unlock();
 890 }
 891
 892 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 893                                                   struct base_data *prev_base)
 894 {
 895         if (base->ref_last == -1 && base->ofs_last == -1) {
 896                 union delta_base base_spec;
 897
 898                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 899                 find_delta_children(&base_spec,
 900                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 901
 902                 memset(&base_spec, 0, sizeof(base_spec));
 903                 base_spec.offset = base->obj->idx.offset;
 904                 find_delta_children(&base_spec,
 905                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 906
 907                 if (base->ref_last == -1 && base->ofs_last == -1) {
 908                         free(base->data);
 909                         return NULL;
 910                 }
 911
 912                 link_base_data(prev_base, base);
 913         }
 914
 915         if (base->ref_first <= base->ref_last) {
 916                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 917                 struct base_data *result = alloc_base_data();
 918
 919                 assert(child->real_type == OBJ_REF_DELTA);
 920                 resolve_delta(child, base, result);
 921                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 922                         free_base_data(base);
 923
 924                 base->ref_first++;
 925                 return result;
 926         }
 927
 928         if (base->ofs_first <= base->ofs_last) {
 929                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 930                 struct base_data *result = alloc_base_data();
 931
 932                 assert(child->real_type == OBJ_OFS_DELTA);
 933                 resolve_delta(child, base, result);
 934                 if (base->ofs_first == base->ofs_last)
 935                         free_base_data(base);
 936
 937                 base->ofs_first++;
 938                 return result;
 939         }
 940
 941         unlink_base_data(base);
 942         return NULL;
 943 }
 944
 945 static void find_unresolved_deltas(struct base_data *base)
 946 {
 947         struct base_data *new_base, *prev_base = NULL;
 948         for (;;) {
 949                 new_base = find_unresolved_deltas_1(base, prev_base);
 950
 951                 if (new_base) {
 952                         prev_base = base;
 953                         base = new_base;
 954                 } else {
 955                         free(base);
 956                         base = prev_base;
 957                         if (!base)
 958                                 return;
 959                         prev_base = base->base;
 960                 }
 961         }
 962 }
 963
 964 static int compare_delta_entry(const void *a, const void *b)
 965 {
 966         const struct delta_entry *delta_a = a;
 967         const struct delta_entry *delta_b = b;
 968
 969         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 970         return compare_delta_bases(&delta_a->base, &delta_b->base,
 971                                    objects[delta_a->obj_no].type,
 972                                    objects[delta_b->obj_no].type);
 973 }
 974
 975 static void resolve_base(struct object_entry *obj)
 976 {
 977         struct base_data *base_obj = alloc_base_data();
 978         base_obj->obj = obj;
 979         base_obj->data = NULL;
 980         find_unresolved_deltas(base_obj);
 981 }
 982
 983 #ifndef NO_PTHREADS
 984 static void *threaded_second_pass(void *data)
 985 {
 986         set_thread_data(data);
 987         for (;;) {
 988                 int i;
 989                 counter_lock();
 990                 display_progress(progress, nr_resolved_deltas);
 991                 counter_unlock();
 992                 work_lock();
 993                 while (nr_dispatched < nr_objects &&
 994                        is_delta_type(objects[nr_dispatched].type))
 995                         nr_dispatched++;
 996                 if (nr_dispatched >= nr_objects) {
 997                         work_unlock();
 998                         break;
 999                 }
1000                 i = nr_dispatched++;
1001                 work_unlock();
1002
1003                 resolve_base(&objects[i]);
1004         }
1005         return NULL;
1006 }
1007 #endif
1008
1009 /*
1010  * First pass:
1011  * - find locations of all objects;
1012  * - calculate SHA1 of all non-delta objects;
1013  * - remember base (SHA1 or offset) for all deltas.
1014  */
1015 static void parse_pack_objects(unsigned char *sha1)
1016 {
1017         int i, nr_delays = 0;
1018         struct delta_entry *delta = deltas;
1019         struct stat st;
1020
1021         if (verbose)
1022                 progress = start_progress(
1023                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1024                                 nr_objects);
1025         for (i = 0; i < nr_objects; i++) {
1026                 struct object_entry *obj = &objects[i];
1027                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1028                 obj->real_type = obj->type;
1029                 if (is_delta_type(obj->type)) {
1030                         nr_deltas++;
1031                         delta->obj_no = i;
1032                         delta++;
1033                 } else if (!data) {
1034                         /* large blobs, check later */
1035                         obj->real_type = OBJ_BAD;
1036                         nr_delays++;
1037                 } else
1038                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1039                 free(data);
1040                 display_progress(progress, i+1);
1041         }
1042         objects[i].idx.offset = consumed_bytes;
1043         stop_progress(&progress);
1044
1045         /* Check pack integrity */
1046         flush();
1047         git_SHA1_Final(sha1, &input_ctx);
1048         if (hashcmp(fill(20), sha1))
1049                 die(_("pack is corrupted (SHA1 mismatch)"));
1050         use(20);
1051
1052         /* If input_fd is a file, we should have reached its end now. */
1053         if (fstat(input_fd, &st))
1054                 die_errno(_("cannot fstat packfile"));
1055         if (S_ISREG(st.st_mode) &&
1056                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1057                 die(_("pack has junk at the end"));
1058
1059         for (i = 0; i < nr_objects; i++) {
1060                 struct object_entry *obj = &objects[i];
1061                 if (obj->real_type != OBJ_BAD)
1062                         continue;
1063                 obj->real_type = obj->type;
1064                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1065                 nr_delays--;
1066         }
1067         if (nr_delays)
1068                 die(_("confusion beyond insanity in parse_pack_objects()"));
1069 }
1070
1071 /*
1072  * Second pass:
1073  * - for all non-delta objects, look if it is used as a base for
1074  *   deltas;
1075  * - if used as a base, uncompress the object and apply all deltas,
1076  *   recursively checking if the resulting object is used as a base
1077  *   for some more deltas.
1078  */
1079 static void resolve_deltas(void)
1080 {
1081         int i;
1082
1083         if (!nr_deltas)
1084                 return;
1085
1086         /* Sort deltas by base SHA1/offset for fast searching */
1087         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1088               compare_delta_entry);
1089
1090         if (verbose)
1091                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1092
1093 #ifndef NO_PTHREADS
1094         nr_dispatched = 0;
1095         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1096                 init_thread();
1097                 for (i = 0; i < nr_threads; i++) {
1098                         int ret = pthread_create(&thread_data[i].thread, NULL,
1099                                                  threaded_second_pass, thread_data + i);
1100                         if (ret)
1101                                 die(_("unable to create thread: %s"),
1102                                     strerror(ret));
1103                 }
1104                 for (i = 0; i < nr_threads; i++)
1105                         pthread_join(thread_data[i].thread, NULL);
1106                 cleanup_thread();
1107                 return;
1108         }
1109 #endif
1110
1111         for (i = 0; i < nr_objects; i++) {
1112                 struct object_entry *obj = &objects[i];
1113
1114                 if (is_delta_type(obj->type))
1115                         continue;
1116                 resolve_base(obj);
1117                 display_progress(progress, nr_resolved_deltas);
1118         }
1119 }
1120
1121 /*
1122  * Third pass:
1123  * - append objects to convert thin pack to full pack if required
1124  * - write the final 20-byte SHA-1
1125  */
1126 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1127 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1128 {
1129         if (nr_deltas == nr_resolved_deltas) {
1130                 stop_progress(&progress);
1131                 /* Flush remaining pack final 20-byte SHA1. */
1132                 flush();
1133                 return;
1134         }
1135
1136         if (fix_thin_pack) {
1137                 struct sha1file *f;
1138                 unsigned char read_sha1[20], tail_sha1[20];
1139                 struct strbuf msg = STRBUF_INIT;
1140                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1141                 int nr_objects_initial = nr_objects;
1142                 if (nr_unresolved <= 0)
1143                         die(_("confusion beyond insanity"));
1144                 objects = xrealloc(objects,
1145                                    (nr_objects + nr_unresolved + 1)
1146                                    * sizeof(*objects));
1147                 memset(objects + nr_objects + 1, 0,
1148                        nr_unresolved * sizeof(*objects));
1149                 f = sha1fd(output_fd, curr_pack);
1150                 fix_unresolved_deltas(f, nr_unresolved);
1151                 strbuf_addf(&msg, _("completed with %d local objects"),
1152                             nr_objects - nr_objects_initial);
1153                 stop_progress_msg(&progress, msg.buf);
1154                 strbuf_release(&msg);
1155                 sha1close(f, tail_sha1, 0);
1156                 hashcpy(read_sha1, pack_sha1);
1157                 fixup_pack_header_footer(output_fd, pack_sha1,
1158                                          curr_pack, nr_objects,
1159                                          read_sha1, consumed_bytes-20);
1160                 if (hashcmp(read_sha1, tail_sha1) != 0)
1161                         die(_("Unexpected tail checksum for %s "
1162                               "(disk corruption?)"), curr_pack);
1163         }
1164         if (nr_deltas != nr_resolved_deltas)
1165                 die(Q_("pack has %d unresolved delta",
1166                        "pack has %d unresolved deltas",
1167                        nr_deltas - nr_resolved_deltas),
1168                     nr_deltas - nr_resolved_deltas);
1169 }
1170
1171 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1172 {
1173         git_zstream stream;
1174         int status;
1175         unsigned char outbuf[4096];
1176
1177         memset(&stream, 0, sizeof(stream));
1178         git_deflate_init(&stream, zlib_compression_level);
1179         stream.next_in = in;
1180         stream.avail_in = size;
1181
1182         do {
1183                 stream.next_out = outbuf;
1184                 stream.avail_out = sizeof(outbuf);
1185                 status = git_deflate(&stream, Z_FINISH);
1186                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1187         } while (status == Z_OK);
1188
1189         if (status != Z_STREAM_END)
1190                 die(_("unable to deflate appended object (%d)"), status);
1191         size = stream.total_out;
1192         git_deflate_end(&stream);
1193         return size;
1194 }
1195
1196 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1197                                const unsigned char *sha1, void *buf,
1198                                unsigned long size, enum object_type type)
1199 {
1200         struct object_entry *obj = &objects[nr_objects++];
1201         unsigned char header[10];
1202         unsigned long s = size;
1203         int n = 0;
1204         unsigned char c = (type << 4) | (s & 15);
1205         s >>= 4;
1206         while (s) {
1207                 header[n++] = c | 0x80;
1208                 c = s & 0x7f;
1209                 s >>= 7;
1210         }
1211         header[n++] = c;
1212         crc32_begin(f);
1213         sha1write(f, header, n);
1214         obj[0].size = size;
1215         obj[0].hdr_size = n;
1216         obj[0].type = type;
1217         obj[0].real_type = type;
1218         obj[1].idx.offset = obj[0].idx.offset + n;
1219         obj[1].idx.offset += write_compressed(f, buf, size);
1220         obj[0].idx.crc32 = crc32_end(f);
1221         sha1flush(f);
1222         hashcpy(obj->idx.sha1, sha1);
1223         return obj;
1224 }
1225
1226 static int delta_pos_compare(const void *_a, const void *_b)
1227 {
1228         struct delta_entry *a = *(struct delta_entry **)_a;
1229         struct delta_entry *b = *(struct delta_entry **)_b;
1230         return a->obj_no - b->obj_no;
1231 }
1232
1233 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1234 {
1235         struct delta_entry **sorted_by_pos;
1236         int i, n = 0;
1237
1238         /*
1239          * Since many unresolved deltas may well be themselves base objects
1240          * for more unresolved deltas, we really want to include the
1241          * smallest number of base objects that would cover as much delta
1242          * as possible by picking the
1243          * trunc deltas first, allowing for other deltas to resolve without
1244          * additional base objects.  Since most base objects are to be found
1245          * before deltas depending on them, a good heuristic is to start
1246          * resolving deltas in the same order as their position in the pack.
1247          */
1248         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1249         for (i = 0; i < nr_deltas; i++) {
1250                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1251                         continue;
1252                 sorted_by_pos[n++] = &deltas[i];
1253         }
1254         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1255
1256         for (i = 0; i < n; i++) {
1257                 struct delta_entry *d = sorted_by_pos[i];
1258                 enum object_type type;
1259                 struct base_data *base_obj = alloc_base_data();
1260
1261                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1262                         continue;
1263                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1264                 if (!base_obj->data)
1265                         continue;
1266
1267                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1268                                 base_obj->size, typename(type)))
1269                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1270                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1271                                         base_obj->data, base_obj->size, type);
1272                 find_unresolved_deltas(base_obj);
1273                 display_progress(progress, nr_resolved_deltas);
1274         }
1275         free(sorted_by_pos);
1276 }
1277
1278 static void final(const char *final_pack_name, const char *curr_pack_name,
1279                   const char *final_index_name, const char *curr_index_name,
1280                   const char *keep_name, const char *keep_msg,
1281                   unsigned char *sha1)
1282 {
1283         const char *report = "pack";
1284         char name[PATH_MAX];
1285         int err;
1286
1287         if (!from_stdin) {
1288                 close(input_fd);
1289         } else {
1290                 fsync_or_die(output_fd, curr_pack_name);
1291                 err = close(output_fd);
1292                 if (err)
1293                         die_errno(_("error while closing pack file"));
1294         }
1295
1296         if (keep_msg) {
1297                 int keep_fd, keep_msg_len = strlen(keep_msg);
1298
1299                 if (!keep_name)
1300                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1301                 else
1302                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1303
1304                 if (keep_fd < 0) {
1305                         if (errno != EEXIST)
1306                                 die_errno(_("cannot write keep file '%s'"),
1307                                           keep_name ? keep_name : name);
1308                 } else {
1309                         if (keep_msg_len > 0) {
1310                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1311                                 write_or_die(keep_fd, "\n", 1);
1312                         }
1313                         if (close(keep_fd) != 0)
1314                                 die_errno(_("cannot close written keep file '%s'"),
1315                                           keep_name ? keep_name : name);
1316                         report = "keep";
1317                 }
1318         }
1319
1320         if (final_pack_name != curr_pack_name) {
1321                 if (!final_pack_name) {
1322                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1323                                  get_object_directory(), sha1_to_hex(sha1));
1324                         final_pack_name = name;
1325                 }
1326                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1327                         die(_("cannot store pack file"));
1328         } else if (from_stdin)
1329                 chmod(final_pack_name, 0444);
1330
1331         if (final_index_name != curr_index_name) {
1332                 if (!final_index_name) {
1333                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1334                                  get_object_directory(), sha1_to_hex(sha1));
1335                         final_index_name = name;
1336                 }
1337                 if (move_temp_to_file(curr_index_name, final_index_name))
1338                         die(_("cannot store index file"));
1339         } else
1340                 chmod(final_index_name, 0444);
1341
1342         if (!from_stdin) {
1343                 printf("%s\n", sha1_to_hex(sha1));
1344         } else {
1345                 char buf[48];
1346                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1347                                    report, sha1_to_hex(sha1));
1348                 write_or_die(1, buf, len);
1349
1350                 /*
1351                  * Let's just mimic git-unpack-objects here and write
1352                  * the last part of the input buffer to stdout.
1353                  */
1354                 while (input_len) {
1355                         err = xwrite(1, input_buffer + input_offset, input_len);
1356                         if (err <= 0)
1357                                 break;
1358                         input_len -= err;
1359                         input_offset += err;
1360                 }
1361         }
1362 }
1363
1364 static int git_index_pack_config(const char *k, const char *v, void *cb)
1365 {
1366         struct pack_idx_option *opts = cb;
1367
1368         if (!strcmp(k, "pack.indexversion")) {
1369                 opts->version = git_config_int(k, v);
1370                 if (opts->version > 2)
1371                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1372                 return 0;
1373         }
1374         if (!strcmp(k, "pack.threads")) {
1375                 nr_threads = git_config_int(k, v);
1376                 if (nr_threads < 0)
1377                         die(_("invalid number of threads specified (%d)"),
1378                             nr_threads);
1379 #ifdef NO_PTHREADS
1380                 if (nr_threads != 1)
1381                         warning(_("no threads support, ignoring %s"), k);
1382                 nr_threads = 1;
1383 #endif
1384                 return 0;
1385         }
1386         return git_default_config(k, v, cb);
1387 }
1388
1389 static int cmp_uint32(const void *a_, const void *b_)
1390 {
1391         uint32_t a = *((uint32_t *)a_);
1392         uint32_t b = *((uint32_t *)b_);
1393
1394         return (a < b) ? -1 : (a != b);
1395 }
1396
1397 static void read_v2_anomalous_offsets(struct packed_git *p,
1398                                       struct pack_idx_option *opts)
1399 {
1400         const uint32_t *idx1, *idx2;
1401         uint32_t i;
1402
1403         /* The address of the 4-byte offset table */
1404         idx1 = (((const uint32_t *)p->index_data)
1405                 + 2 /* 8-byte header */
1406                 + 256 /* fan out */
1407                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1408                 + p->num_objects /* CRC32 table */
1409                 );
1410
1411         /* The address of the 8-byte offset table */
1412         idx2 = idx1 + p->num_objects;
1413
1414         for (i = 0; i < p->num_objects; i++) {
1415                 uint32_t off = ntohl(idx1[i]);
1416                 if (!(off & 0x80000000))
1417                         continue;
1418                 off = off & 0x7fffffff;
1419                 if (idx2[off * 2])
1420                         continue;
1421                 /*
1422                  * The real offset is ntohl(idx2[off * 2]) in high 4
1423                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1424                  * octets.  But idx2[off * 2] is Zero!!!
1425                  */
1426                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1427                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1428         }
1429
1430         if (1 < opts->anomaly_nr)
1431                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1432 }
1433
1434 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1435 {
1436         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1437
1438         if (!p)
1439                 die(_("Cannot open existing pack file '%s'"), pack_name);
1440         if (open_pack_index(p))
1441                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1442
1443         /* Read the attributes from the existing idx file */
1444         opts->version = p->index_version;
1445
1446         if (opts->version == 2)
1447                 read_v2_anomalous_offsets(p, opts);
1448
1449         /*
1450          * Get rid of the idx file as we do not need it anymore.
1451          * NEEDSWORK: extract this bit from free_pack_by_name() in
1452          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1453          * know we haven't installed this pack (hence we never have
1454          * read anything from it).
1455          */
1456         close_pack_index(p);
1457         free(p);
1458 }
1459
1460 static void show_pack_info(int stat_only)
1461 {
1462         int i, baseobjects = nr_objects - nr_deltas;
1463         unsigned long *chain_histogram = NULL;
1464
1465         if (deepest_delta)
1466                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1467
1468         for (i = 0; i < nr_objects; i++) {
1469                 struct object_entry *obj = &objects[i];
1470
1471                 if (is_delta_type(obj->type))
1472                         chain_histogram[obj->delta_depth - 1]++;
1473                 if (stat_only)
1474                         continue;
1475                 printf("%s %-6s %lu %lu %"PRIuMAX,
1476                        sha1_to_hex(obj->idx.sha1),
1477                        typename(obj->real_type), obj->size,
1478                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1479                        (uintmax_t)obj->idx.offset);
1480                 if (is_delta_type(obj->type)) {
1481                         struct object_entry *bobj = &objects[obj->base_object_no];
1482                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1483                 }
1484                 putchar('\n');
1485         }
1486
1487         if (baseobjects)
1488                 printf_ln(Q_("non delta: %d object",
1489                              "non delta: %d objects",
1490                              baseobjects),
1491                           baseobjects);
1492         for (i = 0; i < deepest_delta; i++) {
1493                 if (!chain_histogram[i])
1494                         continue;
1495                 printf_ln(Q_("chain length = %d: %lu object",
1496                              "chain length = %d: %lu objects",
1497                              chain_histogram[i]),
1498                           i + 1,
1499                           chain_histogram[i]);
1500         }
1501 }
1502
1503 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1504 {
1505         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1506         const char *curr_index;
1507         const char *index_name = NULL, *pack_name = NULL;
1508         const char *keep_name = NULL, *keep_msg = NULL;
1509         char *index_name_buf = NULL, *keep_name_buf = NULL;
1510         struct pack_idx_entry **idx_objects;
1511         struct pack_idx_option opts;
1512         unsigned char pack_sha1[20];
1513         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1514
1515         if (argc == 2 && !strcmp(argv[1], "-h"))
1516                 usage(index_pack_usage);
1517
1518         check_replace_refs = 0;
1519
1520         reset_pack_idx_option(&opts);
1521         git_config(git_index_pack_config, &opts);
1522         if (prefix && chdir(prefix))
1523                 die(_("Cannot come back to cwd"));
1524
1525         for (i = 1; i < argc; i++) {
1526                 const char *arg = argv[i];
1527
1528                 if (*arg == '-') {
1529                         if (!strcmp(arg, "--stdin")) {
1530                                 from_stdin = 1;
1531                         } else if (!strcmp(arg, "--fix-thin")) {
1532                                 fix_thin_pack = 1;
1533                         } else if (!strcmp(arg, "--strict")) {
1534                                 strict = 1;
1535                                 do_fsck_object = 1;
1536                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1537                                 strict = 1;
1538                                 check_self_contained_and_connected = 1;
1539                         } else if (!strcmp(arg, "--verify")) {
1540                                 verify = 1;
1541                         } else if (!strcmp(arg, "--verify-stat")) {
1542                                 verify = 1;
1543                                 show_stat = 1;
1544                         } else if (!strcmp(arg, "--verify-stat-only")) {
1545                                 verify = 1;
1546                                 show_stat = 1;
1547                                 stat_only = 1;
1548                         } else if (!strcmp(arg, "--keep")) {
1549                                 keep_msg = "";
1550                         } else if (starts_with(arg, "--keep=")) {
1551                                 keep_msg = arg + 7;
1552                         } else if (starts_with(arg, "--threads=")) {
1553                                 char *end;
1554                                 nr_threads = strtoul(arg+10, &end, 0);
1555                                 if (!arg[10] || *end || nr_threads < 0)
1556                                         usage(index_pack_usage);
1557 #ifdef NO_PTHREADS
1558                                 if (nr_threads != 1)
1559                                         warning(_("no threads support, "
1560                                                   "ignoring %s"), arg);
1561                                 nr_threads = 1;
1562 #endif
1563                         } else if (starts_with(arg, "--pack_header=")) {
1564                                 struct pack_header *hdr;
1565                                 char *c;
1566
1567                                 hdr = (struct pack_header *)input_buffer;
1568                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1569                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1570                                 if (*c != ',')
1571                                         die(_("bad %s"), arg);
1572                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1573                                 if (*c)
1574                                         die(_("bad %s"), arg);
1575                                 input_len = sizeof(*hdr);
1576                         } else if (!strcmp(arg, "-v")) {
1577                                 verbose = 1;
1578                         } else if (!strcmp(arg, "-o")) {
1579                                 if (index_name || (i+1) >= argc)
1580                                         usage(index_pack_usage);
1581                                 index_name = argv[++i];
1582                         } else if (starts_with(arg, "--index-version=")) {
1583                                 char *c;
1584                                 opts.version = strtoul(arg + 16, &c, 10);
1585                                 if (opts.version > 2)
1586                                         die(_("bad %s"), arg);
1587                                 if (*c == ',')
1588                                         opts.off32_limit = strtoul(c+1, &c, 0);
1589                                 if (*c || opts.off32_limit & 0x80000000)
1590                                         die(_("bad %s"), arg);
1591                         } else
1592                                 usage(index_pack_usage);
1593                         continue;
1594                 }
1595
1596                 if (pack_name)
1597                         usage(index_pack_usage);
1598                 pack_name = arg;
1599         }
1600
1601         if (!pack_name && !from_stdin)
1602                 usage(index_pack_usage);
1603         if (fix_thin_pack && !from_stdin)
1604                 die(_("--fix-thin cannot be used without --stdin"));
1605         if (!index_name && pack_name) {
1606                 int len = strlen(pack_name);
1607                 if (!has_extension(pack_name, ".pack"))
1608                         die(_("packfile name '%s' does not end with '.pack'"),
1609                             pack_name);
1610                 index_name_buf = xmalloc(len);
1611                 memcpy(index_name_buf, pack_name, len - 5);
1612                 strcpy(index_name_buf + len - 5, ".idx");
1613                 index_name = index_name_buf;
1614         }
1615         if (keep_msg && !keep_name && pack_name) {
1616                 int len = strlen(pack_name);
1617                 if (!has_extension(pack_name, ".pack"))
1618                         die(_("packfile name '%s' does not end with '.pack'"),
1619                             pack_name);
1620                 keep_name_buf = xmalloc(len);
1621                 memcpy(keep_name_buf, pack_name, len - 5);
1622                 strcpy(keep_name_buf + len - 5, ".keep");
1623                 keep_name = keep_name_buf;
1624         }
1625         if (verify) {
1626                 if (!index_name)
1627                         die(_("--verify with no packfile name given"));
1628                 read_idx_option(&opts, index_name);
1629                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1630         }
1631         if (strict)
1632                 opts.flags |= WRITE_IDX_STRICT;
1633
1634 #ifndef NO_PTHREADS
1635         if (!nr_threads) {
1636                 nr_threads = online_cpus();
1637                 /* An experiment showed that more threads does not mean faster */
1638                 if (nr_threads > 3)
1639                         nr_threads = 3;
1640         }
1641 #endif
1642
1643         curr_pack = open_pack_file(pack_name);
1644         parse_pack_header();
1645         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1646         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1647         parse_pack_objects(pack_sha1);
1648         resolve_deltas();
1649         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1650         free(deltas);
1651         if (strict)
1652                 foreign_nr = check_objects();
1653
1654         if (show_stat)
1655                 show_pack_info(stat_only);
1656
1657         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1658         for (i = 0; i < nr_objects; i++)
1659                 idx_objects[i] = &objects[i].idx;
1660         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1661         free(idx_objects);
1662
1663         if (!verify)
1664                 final(pack_name, curr_pack,
1665                       index_name, curr_index,
1666                       keep_name, keep_msg,
1667                       pack_sha1);
1668         else
1669                 close(input_fd);
1670         free(objects);
1671         free(index_name_buf);
1672         free(keep_name_buf);
1673         if (pack_name == NULL)
1674                 free((void *) curr_pack);
1675         if (index_name == NULL)
1676                 free((void *) curr_index);
1677
1678         /*
1679          * Let the caller know this pack is not self contained
1680          */
1681         if (check_self_contained_and_connected && foreign_nr)
1682                 return 1;
1683
1684         return 0;
1685 }