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

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned int hdr_size;
  22         enum object_type type;
  23         enum object_type real_type;
  24         unsigned delta_depth;
  25         int base_object_no;
  26 };
  27
  28 union delta_base {
  29         unsigned char sha1[20];
  30         off_t offset;
  31 };
  32
  33 struct base_data {
  34         struct base_data *base;
  35         struct base_data *child;
  36         struct object_entry *obj;
  37         void *data;
  38         unsigned long size;
  39         int ref_first, ref_last;
  40         int ofs_first, ofs_last;
  41 };
  42
  43 struct thread_local {
  44 #ifndef NO_PTHREADS
  45         pthread_t thread;
  46 #endif
  47         struct base_data *base_cache;
  48         size_t base_cache_used;
  49         int pack_fd;
  50 };
  51
  52 /*
  53  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  54  * to memcmp() only the first 20 bytes.
  55  */
  56 #define UNION_BASE_SZ   20
  57
  58 #define FLAG_LINK (1u<<20)
  59 #define FLAG_CHECKED (1u<<21)
  60
  61 struct delta_entry {
  62         union delta_base base;
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct delta_entry *deltas;
  68 static struct thread_local nothread_data;
  69 static int nr_objects;
  70 static int nr_deltas;
  71 static int nr_resolved_deltas;
  72 static int nr_threads;
  73
  74 static int from_stdin;
  75 static int strict;
  76 static int do_fsck_object;
  77 static int verbose;
  78 static int show_stat;
  79 static int check_self_contained_and_connected;
  80
  81 static struct progress *progress;
  82
  83 /* We always read in 4kB chunks. */
  84 static unsigned char input_buffer[4096];
  85 static unsigned int input_offset, input_len;
  86 static off_t consumed_bytes;
  87 static unsigned deepest_delta;
  88 static git_SHA_CTX input_ctx;
  89 static uint32_t input_crc32;
  90 static int input_fd, output_fd;
  91 static const char *curr_pack;
  92
  93 #ifndef NO_PTHREADS
  94
  95 static struct thread_local *thread_data;
  96 static int nr_dispatched;
  97 static int threads_active;
  98
  99 static pthread_mutex_t read_mutex;
 100 #define read_lock()             lock_mutex(&read_mutex)
 101 #define read_unlock()           unlock_mutex(&read_mutex)
 102
 103 static pthread_mutex_t counter_mutex;
 104 #define counter_lock()          lock_mutex(&counter_mutex)
 105 #define counter_unlock()        unlock_mutex(&counter_mutex)
 106
 107 static pthread_mutex_t work_mutex;
 108 #define work_lock()             lock_mutex(&work_mutex)
 109 #define work_unlock()           unlock_mutex(&work_mutex)
 110
 111 static pthread_mutex_t deepest_delta_mutex;
 112 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 113 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 114
 115 static pthread_mutex_t type_cas_mutex;
 116 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 117 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 118
 119 static pthread_key_t key;
 120
 121 static inline void lock_mutex(pthread_mutex_t *mutex)
 122 {
 123         if (threads_active)
 124                 pthread_mutex_lock(mutex);
 125 }
 126
 127 static inline void unlock_mutex(pthread_mutex_t *mutex)
 128 {
 129         if (threads_active)
 130                 pthread_mutex_unlock(mutex);
 131 }
 132
 133 /*
 134  * Mutex and conditional variable can't be statically-initialized on Windows.
 135  */
 136 static void init_thread(void)
 137 {
 138         int i;
 139         init_recursive_mutex(&read_mutex);
 140         pthread_mutex_init(&counter_mutex, NULL);
 141         pthread_mutex_init(&work_mutex, NULL);
 142         pthread_mutex_init(&type_cas_mutex, NULL);
 143         if (show_stat)
 144                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 145         pthread_key_create(&key, NULL);
 146         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 147         for (i = 0; i < nr_threads; i++) {
 148                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 149                 if (thread_data[i].pack_fd == -1)
 150                         die_errno(_("unable to open %s"), curr_pack);
 151         }
 152
 153         threads_active = 1;
 154 }
 155
 156 static void cleanup_thread(void)
 157 {
 158         int i;
 159         if (!threads_active)
 160                 return;
 161         threads_active = 0;
 162         pthread_mutex_destroy(&read_mutex);
 163         pthread_mutex_destroy(&counter_mutex);
 164         pthread_mutex_destroy(&work_mutex);
 165         pthread_mutex_destroy(&type_cas_mutex);
 166         if (show_stat)
 167                 pthread_mutex_destroy(&deepest_delta_mutex);
 168         for (i = 0; i < nr_threads; i++)
 169                 close(thread_data[i].pack_fd);
 170         pthread_key_delete(key);
 171         free(thread_data);
 172 }
 173
 174 #else
 175
 176 #define read_lock()
 177 #define read_unlock()
 178
 179 #define counter_lock()
 180 #define counter_unlock()
 181
 182 #define work_lock()
 183 #define work_unlock()
 184
 185 #define deepest_delta_lock()
 186 #define deepest_delta_unlock()
 187
 188 #define type_cas_lock()
 189 #define type_cas_unlock()
 190
 191 #endif
 192
 193
 194 static int mark_link(struct object *obj, int type, void *data)
 195 {
 196         if (!obj)
 197                 return -1;
 198
 199         if (type != OBJ_ANY && obj->type != type)
 200                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 201
 202         obj->flags |= FLAG_LINK;
 203         return 0;
 204 }
 205
 206 /* The content of each linked object must have been checked
 207    or it must be already present in the object database */
 208 static unsigned check_object(struct object *obj)
 209 {
 210         if (!obj)
 211                 return 0;
 212
 213         if (!(obj->flags & FLAG_LINK))
 214                 return 0;
 215
 216         if (!(obj->flags & FLAG_CHECKED)) {
 217                 unsigned long size;
 218                 int type = sha1_object_info(obj->sha1, &size);
 219                 if (type <= 0)
 220                         die(_("did not receive expected object %s"),
 221                               sha1_to_hex(obj->sha1));
 222                 if (type != obj->type)
 223                         die(_("object %s: expected type %s, found %s"),
 224                             sha1_to_hex(obj->sha1),
 225                             typename(obj->type), typename(type));
 226                 obj->flags |= FLAG_CHECKED;
 227                 return 1;
 228         }
 229
 230         return 0;
 231 }
 232
 233 static unsigned check_objects(void)
 234 {
 235         unsigned i, max, foreign_nr = 0;
 236
 237         max = get_max_object_index();
 238         for (i = 0; i < max; i++)
 239                 foreign_nr += check_object(get_indexed_object(i));
 240         return foreign_nr;
 241 }
 242
 243
 244 /* Discard current buffer used content. */
 245 static void flush(void)
 246 {
 247         if (input_offset) {
 248                 if (output_fd >= 0)
 249                         write_or_die(output_fd, input_buffer, input_offset);
 250                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 251                 memmove(input_buffer, input_buffer + input_offset, input_len);
 252                 input_offset = 0;
 253         }
 254 }
 255
 256 /*
 257  * Make sure at least "min" bytes are available in the buffer, and
 258  * return the pointer to the buffer.
 259  */
 260 static void *fill(int min)
 261 {
 262         if (min <= input_len)
 263                 return input_buffer + input_offset;
 264         if (min > sizeof(input_buffer))
 265                 die(Q_("cannot fill %d byte",
 266                        "cannot fill %d bytes",
 267                        min),
 268                     min);
 269         flush();
 270         do {
 271                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 272                                 sizeof(input_buffer) - input_len);
 273                 if (ret <= 0) {
 274                         if (!ret)
 275                                 die(_("early EOF"));
 276                         die_errno(_("read error on input"));
 277                 }
 278                 input_len += ret;
 279                 if (from_stdin)
 280                         display_throughput(progress, consumed_bytes + input_len);
 281         } while (input_len < min);
 282         return input_buffer;
 283 }
 284
 285 static void use(int bytes)
 286 {
 287         if (bytes > input_len)
 288                 die(_("used more bytes than were available"));
 289         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 290         input_len -= bytes;
 291         input_offset += bytes;
 292
 293         /* make sure off_t is sufficiently large not to wrap */
 294         if (signed_add_overflows(consumed_bytes, bytes))
 295                 die(_("pack too large for current definition of off_t"));
 296         consumed_bytes += bytes;
 297 }
 298
 299 static const char *open_pack_file(const char *pack_name)
 300 {
 301         if (from_stdin) {
 302                 input_fd = 0;
 303                 if (!pack_name) {
 304                         static char tmp_file[PATH_MAX];
 305                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 306                                                 "pack/tmp_pack_XXXXXX");
 307                         pack_name = xstrdup(tmp_file);
 308                 } else
 309                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 310                 if (output_fd < 0)
 311                         die_errno(_("unable to create '%s'"), pack_name);
 312                 nothread_data.pack_fd = output_fd;
 313         } else {
 314                 input_fd = open(pack_name, O_RDONLY);
 315                 if (input_fd < 0)
 316                         die_errno(_("cannot open packfile '%s'"), pack_name);
 317                 output_fd = -1;
 318                 nothread_data.pack_fd = input_fd;
 319         }
 320         git_SHA1_Init(&input_ctx);
 321         return pack_name;
 322 }
 323
 324 static void parse_pack_header(void)
 325 {
 326         struct pack_header *hdr = fill(sizeof(struct pack_header));
 327
 328         /* Header consistency check */
 329         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 330                 die(_("pack signature mismatch"));
 331         if (!pack_version_ok(hdr->hdr_version))
 332                 die(_("pack version %"PRIu32" unsupported"),
 333                         ntohl(hdr->hdr_version));
 334
 335         nr_objects = ntohl(hdr->hdr_entries);
 336         use(sizeof(struct pack_header));
 337 }
 338
 339 static NORETURN void bad_object(unsigned long offset, const char *format,
 340                        ...) __attribute__((format (printf, 2, 3)));
 341
 342 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 343 {
 344         va_list params;
 345         char buf[1024];
 346
 347         va_start(params, format);
 348         vsnprintf(buf, sizeof(buf), format, params);
 349         va_end(params);
 350         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 351 }
 352
 353 static inline struct thread_local *get_thread_data(void)
 354 {
 355 #ifndef NO_PTHREADS
 356         if (threads_active)
 357                 return pthread_getspecific(key);
 358         assert(!threads_active &&
 359                "This should only be reached when all threads are gone");
 360 #endif
 361         return &nothread_data;
 362 }
 363
 364 #ifndef NO_PTHREADS
 365 static void set_thread_data(struct thread_local *data)
 366 {
 367         if (threads_active)
 368                 pthread_setspecific(key, data);
 369 }
 370 #endif
 371
 372 static struct base_data *alloc_base_data(void)
 373 {
 374         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 375         base->ref_last = -1;
 376         base->ofs_last = -1;
 377         return base;
 378 }
 379
 380 static void free_base_data(struct base_data *c)
 381 {
 382         if (c->data) {
 383                 free(c->data);
 384                 c->data = NULL;
 385                 get_thread_data()->base_cache_used -= c->size;
 386         }
 387 }
 388
 389 static void prune_base_data(struct base_data *retain)
 390 {
 391         struct base_data *b;
 392         struct thread_local *data = get_thread_data();
 393         for (b = data->base_cache;
 394              data->base_cache_used > delta_base_cache_limit && b;
 395              b = b->child) {
 396                 if (b->data && b != retain)
 397                         free_base_data(b);
 398         }
 399 }
 400
 401 static void link_base_data(struct base_data *base, struct base_data *c)
 402 {
 403         if (base)
 404                 base->child = c;
 405         else
 406                 get_thread_data()->base_cache = c;
 407
 408         c->base = base;
 409         c->child = NULL;
 410         if (c->data)
 411                 get_thread_data()->base_cache_used += c->size;
 412         prune_base_data(c);
 413 }
 414
 415 static void unlink_base_data(struct base_data *c)
 416 {
 417         struct base_data *base = c->base;
 418         if (base)
 419                 base->child = NULL;
 420         else
 421                 get_thread_data()->base_cache = NULL;
 422         free_base_data(c);
 423 }
 424
 425 static int is_delta_type(enum object_type type)
 426 {
 427         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 428 }
 429
 430 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 431                                enum object_type type, unsigned char *sha1)
 432 {
 433         static char fixed_buf[8192];
 434         int status;
 435         git_zstream stream;
 436         void *buf;
 437         git_SHA_CTX c;
 438         char hdr[32];
 439         int hdrlen;
 440
 441         if (!is_delta_type(type)) {
 442                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 443                 git_SHA1_Init(&c);
 444                 git_SHA1_Update(&c, hdr, hdrlen);
 445         } else
 446                 sha1 = NULL;
 447         if (type == OBJ_BLOB && size > big_file_threshold)
 448                 buf = fixed_buf;
 449         else
 450                 buf = xmallocz(size);
 451
 452         memset(&stream, 0, sizeof(stream));
 453         git_inflate_init(&stream);
 454         stream.next_out = buf;
 455         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 456
 457         do {
 458                 unsigned char *last_out = stream.next_out;
 459                 stream.next_in = fill(1);
 460                 stream.avail_in = input_len;
 461                 status = git_inflate(&stream, 0);
 462                 use(input_len - stream.avail_in);
 463                 if (sha1)
 464                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 465                 if (buf == fixed_buf) {
 466                         stream.next_out = buf;
 467                         stream.avail_out = sizeof(fixed_buf);
 468                 }
 469         } while (status == Z_OK);
 470         if (stream.total_out != size || status != Z_STREAM_END)
 471                 bad_object(offset, _("inflate returned %d"), status);
 472         git_inflate_end(&stream);
 473         if (sha1)
 474                 git_SHA1_Final(sha1, &c);
 475         return buf == fixed_buf ? NULL : buf;
 476 }
 477
 478 static void *unpack_raw_entry(struct object_entry *obj,
 479                               union delta_base *delta_base,
 480                               unsigned char *sha1)
 481 {
 482         unsigned char *p;
 483         unsigned long size, c;
 484         off_t base_offset;
 485         unsigned shift;
 486         void *data;
 487
 488         obj->idx.offset = consumed_bytes;
 489         input_crc32 = crc32(0, NULL, 0);
 490
 491         p = fill(1);
 492         c = *p;
 493         use(1);
 494         obj->type = (c >> 4) & 7;
 495         size = (c & 15);
 496         shift = 4;
 497         while (c & 0x80) {
 498                 p = fill(1);
 499                 c = *p;
 500                 use(1);
 501                 size += (c & 0x7f) << shift;
 502                 shift += 7;
 503         }
 504         obj->size = size;
 505
 506         switch (obj->type) {
 507         case OBJ_REF_DELTA:
 508                 hashcpy(delta_base->sha1, fill(20));
 509                 use(20);
 510                 break;
 511         case OBJ_OFS_DELTA:
 512                 memset(delta_base, 0, sizeof(*delta_base));
 513                 p = fill(1);
 514                 c = *p;
 515                 use(1);
 516                 base_offset = c & 127;
 517                 while (c & 128) {
 518                         base_offset += 1;
 519                         if (!base_offset || MSB(base_offset, 7))
 520                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 521                         p = fill(1);
 522                         c = *p;
 523                         use(1);
 524                         base_offset = (base_offset << 7) + (c & 127);
 525                 }
 526                 delta_base->offset = obj->idx.offset - base_offset;
 527                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 528                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 529                 break;
 530         case OBJ_COMMIT:
 531         case OBJ_TREE:
 532         case OBJ_BLOB:
 533         case OBJ_TAG:
 534                 break;
 535         default:
 536                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 537         }
 538         obj->hdr_size = consumed_bytes - obj->idx.offset;
 539
 540         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 541         obj->idx.crc32 = input_crc32;
 542         return data;
 543 }
 544
 545 static void *unpack_data(struct object_entry *obj,
 546                          int (*consume)(const unsigned char *, unsigned long, void *),
 547                          void *cb_data)
 548 {
 549         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 550         unsigned long len = obj[1].idx.offset - from;
 551         unsigned char *data, *inbuf;
 552         git_zstream stream;
 553         int status;
 554
 555         data = xmallocz(consume ? 64*1024 : obj->size);
 556         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 557
 558         memset(&stream, 0, sizeof(stream));
 559         git_inflate_init(&stream);
 560         stream.next_out = data;
 561         stream.avail_out = consume ? 64*1024 : obj->size;
 562
 563         do {
 564                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 565                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 566                 if (n < 0)
 567                         die_errno(_("cannot pread pack file"));
 568                 if (!n)
 569                         die(Q_("premature end of pack file, %lu byte missing",
 570                                "premature end of pack file, %lu bytes missing",
 571                                len),
 572                             len);
 573                 from += n;
 574                 len -= n;
 575                 stream.next_in = inbuf;
 576                 stream.avail_in = n;
 577                 if (!consume)
 578                         status = git_inflate(&stream, 0);
 579                 else {
 580                         do {
 581                                 status = git_inflate(&stream, 0);
 582                                 if (consume(data, stream.next_out - data, cb_data)) {
 583                                         free(inbuf);
 584                                         free(data);
 585                                         return NULL;
 586                                 }
 587                                 stream.next_out = data;
 588                                 stream.avail_out = 64*1024;
 589                         } while (status == Z_OK && stream.avail_in);
 590                 }
 591         } while (len && status == Z_OK && !stream.avail_in);
 592
 593         /* This has been inflated OK when first encountered, so... */
 594         if (status != Z_STREAM_END || stream.total_out != obj->size)
 595                 die(_("serious inflate inconsistency"));
 596
 597         git_inflate_end(&stream);
 598         free(inbuf);
 599         if (consume) {
 600                 free(data);
 601                 data = NULL;
 602         }
 603         return data;
 604 }
 605
 606 static void *get_data_from_pack(struct object_entry *obj)
 607 {
 608         return unpack_data(obj, NULL, NULL);
 609 }
 610
 611 static int compare_delta_bases(const union delta_base *base1,
 612                                const union delta_base *base2,
 613                                enum object_type type1,
 614                                enum object_type type2)
 615 {
 616         int cmp = type1 - type2;
 617         if (cmp)
 618                 return cmp;
 619         return memcmp(base1, base2, UNION_BASE_SZ);
 620 }
 621
 622 static int find_delta(const union delta_base *base, enum object_type type)
 623 {
 624         int first = 0, last = nr_deltas;
 625
 626         while (first < last) {
 627                 int next = (first + last) / 2;
 628                 struct delta_entry *delta = &deltas[next];
 629                 int cmp;
 630
 631                 cmp = compare_delta_bases(base, &delta->base,
 632                                           type, objects[delta->obj_no].type);
 633                 if (!cmp)
 634                         return next;
 635                 if (cmp < 0) {
 636                         last = next;
 637                         continue;
 638                 }
 639                 first = next+1;
 640         }
 641         return -first-1;
 642 }
 643
 644 static void find_delta_children(const union delta_base *base,
 645                                 int *first_index, int *last_index,
 646                                 enum object_type type)
 647 {
 648         int first = find_delta(base, type);
 649         int last = first;
 650         int end = nr_deltas - 1;
 651
 652         if (first < 0) {
 653                 *first_index = 0;
 654                 *last_index = -1;
 655                 return;
 656         }
 657         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 658                 --first;
 659         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 660                 ++last;
 661         *first_index = first;
 662         *last_index = last;
 663 }
 664
 665 struct compare_data {
 666         struct object_entry *entry;
 667         struct git_istream *st;
 668         unsigned char *buf;
 669         unsigned long buf_size;
 670 };
 671
 672 static int compare_objects(const unsigned char *buf, unsigned long size,
 673                            void *cb_data)
 674 {
 675         struct compare_data *data = cb_data;
 676
 677         if (data->buf_size < size) {
 678                 free(data->buf);
 679                 data->buf = xmalloc(size);
 680                 data->buf_size = size;
 681         }
 682
 683         while (size) {
 684                 ssize_t len = read_istream(data->st, data->buf, size);
 685                 if (len == 0)
 686                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 687                             sha1_to_hex(data->entry->idx.sha1));
 688                 if (len < 0)
 689                         die(_("unable to read %s"),
 690                             sha1_to_hex(data->entry->idx.sha1));
 691                 if (memcmp(buf, data->buf, len))
 692                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 693                             sha1_to_hex(data->entry->idx.sha1));
 694                 size -= len;
 695                 buf += len;
 696         }
 697         return 0;
 698 }
 699
 700 static int check_collison(struct object_entry *entry)
 701 {
 702         struct compare_data data;
 703         enum object_type type;
 704         unsigned long size;
 705
 706         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 707                 return -1;
 708
 709         memset(&data, 0, sizeof(data));
 710         data.entry = entry;
 711         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 712         if (!data.st)
 713                 return -1;
 714         if (size != entry->size || type != entry->type)
 715                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 716                     sha1_to_hex(entry->idx.sha1));
 717         unpack_data(entry, compare_objects, &data);
 718         close_istream(data.st);
 719         free(data.buf);
 720         return 0;
 721 }
 722
 723 static void sha1_object(const void *data, struct object_entry *obj_entry,
 724                         unsigned long size, enum object_type type,
 725                         const unsigned char *sha1)
 726 {
 727         void *new_data = NULL;
 728         int collision_test_needed;
 729
 730         assert(data || obj_entry);
 731
 732         read_lock();
 733         collision_test_needed = has_sha1_file(sha1);
 734         read_unlock();
 735
 736         if (collision_test_needed && !data) {
 737                 read_lock();
 738                 if (!check_collison(obj_entry))
 739                         collision_test_needed = 0;
 740                 read_unlock();
 741         }
 742         if (collision_test_needed) {
 743                 void *has_data;
 744                 enum object_type has_type;
 745                 unsigned long has_size;
 746                 read_lock();
 747                 has_type = sha1_object_info(sha1, &has_size);
 748                 if (has_type != type || has_size != size)
 749                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 750                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 751                 read_unlock();
 752                 if (!data)
 753                         data = new_data = get_data_from_pack(obj_entry);
 754                 if (!has_data)
 755                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 756                 if (size != has_size || type != has_type ||
 757                     memcmp(data, has_data, size) != 0)
 758                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 759                 free(has_data);
 760         }
 761
 762         if (strict) {
 763                 read_lock();
 764                 if (type == OBJ_BLOB) {
 765                         struct blob *blob = lookup_blob(sha1);
 766                         if (blob)
 767                                 blob->object.flags |= FLAG_CHECKED;
 768                         else
 769                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 770                 } else {
 771                         struct object *obj;
 772                         int eaten;
 773                         void *buf = (void *) data;
 774
 775                         assert(data && "data can only be NULL for large _blobs_");
 776
 777                         /*
 778                          * we do not need to free the memory here, as the
 779                          * buf is deleted by the caller.
 780                          */
 781                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 782                         if (!obj)
 783                                 die(_("invalid %s"), typename(type));
 784                         if (do_fsck_object &&
 785                             fsck_object(obj, buf, size, 1,
 786                                     fsck_error_function))
 787                                 die(_("Error in object"));
 788                         if (fsck_walk(obj, mark_link, NULL))
 789                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 790
 791                         if (obj->type == OBJ_TREE) {
 792                                 struct tree *item = (struct tree *) obj;
 793                                 item->buffer = NULL;
 794                                 obj->parsed = 0;
 795                         }
 796                         if (obj->type == OBJ_COMMIT) {
 797                                 struct commit *commit = (struct commit *) obj;
 798                                 if (detach_commit_buffer(commit, NULL) != data)
 799                                         die("BUG: parse_object_buffer transmogrified our buffer");
 800                         }
 801                         obj->flags |= FLAG_CHECKED;
 802                 }
 803                 read_unlock();
 804         }
 805
 806         free(new_data);
 807 }
 808
 809 /*
 810  * This function is part of find_unresolved_deltas(). There are two
 811  * walkers going in the opposite ways.
 812  *
 813  * The first one in find_unresolved_deltas() traverses down from
 814  * parent node to children, deflating nodes along the way. However,
 815  * memory for deflated nodes is limited by delta_base_cache_limit, so
 816  * at some point parent node's deflated content may be freed.
 817  *
 818  * The second walker is this function, which goes from current node up
 819  * to top parent if necessary to deflate the node. In normal
 820  * situation, its parent node would be already deflated, so it just
 821  * needs to apply delta.
 822  *
 823  * In the worst case scenario, parent node is no longer deflated because
 824  * we're running out of delta_base_cache_limit; we need to re-deflate
 825  * parents, possibly up to the top base.
 826  *
 827  * All deflated objects here are subject to be freed if we exceed
 828  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 829  * just need to make sure the last node is not freed.
 830  */
 831 static void *get_base_data(struct base_data *c)
 832 {
 833         if (!c->data) {
 834                 struct object_entry *obj = c->obj;
 835                 struct base_data **delta = NULL;
 836                 int delta_nr = 0, delta_alloc = 0;
 837
 838                 while (is_delta_type(c->obj->type) && !c->data) {
 839                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 840                         delta[delta_nr++] = c;
 841                         c = c->base;
 842                 }
 843                 if (!delta_nr) {
 844                         c->data = get_data_from_pack(obj);
 845                         c->size = obj->size;
 846                         get_thread_data()->base_cache_used += c->size;
 847                         prune_base_data(c);
 848                 }
 849                 for (; delta_nr > 0; delta_nr--) {
 850                         void *base, *raw;
 851                         c = delta[delta_nr - 1];
 852                         obj = c->obj;
 853                         base = get_base_data(c->base);
 854                         raw = get_data_from_pack(obj);
 855                         c->data = patch_delta(
 856                                 base, c->base->size,
 857                                 raw, obj->size,
 858                                 &c->size);
 859                         free(raw);
 860                         if (!c->data)
 861                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 862                         get_thread_data()->base_cache_used += c->size;
 863                         prune_base_data(c);
 864                 }
 865                 free(delta);
 866         }
 867         return c->data;
 868 }
 869
 870 static void resolve_delta(struct object_entry *delta_obj,
 871                           struct base_data *base, struct base_data *result)
 872 {
 873         void *base_data, *delta_data;
 874
 875         if (show_stat) {
 876                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 877                 deepest_delta_lock();
 878                 if (deepest_delta < delta_obj->delta_depth)
 879                         deepest_delta = delta_obj->delta_depth;
 880                 deepest_delta_unlock();
 881         }
 882         delta_obj->base_object_no = base->obj - objects;
 883         delta_data = get_data_from_pack(delta_obj);
 884         base_data = get_base_data(base);
 885         result->obj = delta_obj;
 886         result->data = patch_delta(base_data, base->size,
 887                                    delta_data, delta_obj->size, &result->size);
 888         free(delta_data);
 889         if (!result->data)
 890                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 891         hash_sha1_file(result->data, result->size,
 892                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 893         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 894                     delta_obj->idx.sha1);
 895         counter_lock();
 896         nr_resolved_deltas++;
 897         counter_unlock();
 898 }
 899
 900 /*
 901  * Standard boolean compare-and-swap: atomically check whether "*type" is
 902  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 903  * and return false.
 904  */
 905 static int compare_and_swap_type(enum object_type *type,
 906                                  enum object_type want,
 907                                  enum object_type set)
 908 {
 909         enum object_type old;
 910
 911         type_cas_lock();
 912         old = *type;
 913         if (old == want)
 914                 *type = set;
 915         type_cas_unlock();
 916
 917         return old == want;
 918 }
 919
 920 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 921                                                   struct base_data *prev_base)
 922 {
 923         if (base->ref_last == -1 && base->ofs_last == -1) {
 924                 union delta_base base_spec;
 925
 926                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 927                 find_delta_children(&base_spec,
 928                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 929
 930                 memset(&base_spec, 0, sizeof(base_spec));
 931                 base_spec.offset = base->obj->idx.offset;
 932                 find_delta_children(&base_spec,
 933                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 934
 935                 if (base->ref_last == -1 && base->ofs_last == -1) {
 936                         free(base->data);
 937                         return NULL;
 938                 }
 939
 940                 link_base_data(prev_base, base);
 941         }
 942
 943         if (base->ref_first <= base->ref_last) {
 944                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 945                 struct base_data *result = alloc_base_data();
 946
 947                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
 948                                            base->obj->real_type))
 949                         die("BUG: child->real_type != OBJ_REF_DELTA");
 950
 951                 resolve_delta(child, base, result);
 952                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 953                         free_base_data(base);
 954
 955                 base->ref_first++;
 956                 return result;
 957         }
 958
 959         if (base->ofs_first <= base->ofs_last) {
 960                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 961                 struct base_data *result = alloc_base_data();
 962
 963                 assert(child->real_type == OBJ_OFS_DELTA);
 964                 child->real_type = base->obj->real_type;
 965                 resolve_delta(child, base, result);
 966                 if (base->ofs_first == base->ofs_last)
 967                         free_base_data(base);
 968
 969                 base->ofs_first++;
 970                 return result;
 971         }
 972
 973         unlink_base_data(base);
 974         return NULL;
 975 }
 976
 977 static void find_unresolved_deltas(struct base_data *base)
 978 {
 979         struct base_data *new_base, *prev_base = NULL;
 980         for (;;) {
 981                 new_base = find_unresolved_deltas_1(base, prev_base);
 982
 983                 if (new_base) {
 984                         prev_base = base;
 985                         base = new_base;
 986                 } else {
 987                         free(base);
 988                         base = prev_base;
 989                         if (!base)
 990                                 return;
 991                         prev_base = base->base;
 992                 }
 993         }
 994 }
 995
 996 static int compare_delta_entry(const void *a, const void *b)
 997 {
 998         const struct delta_entry *delta_a = a;
 999         const struct delta_entry *delta_b = b;
1000
1001         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
1002         return compare_delta_bases(&delta_a->base, &delta_b->base,
1003                                    objects[delta_a->obj_no].type,
1004                                    objects[delta_b->obj_no].type);
1005 }
1006
1007 static void resolve_base(struct object_entry *obj)
1008 {
1009         struct base_data *base_obj = alloc_base_data();
1010         base_obj->obj = obj;
1011         base_obj->data = NULL;
1012         find_unresolved_deltas(base_obj);
1013 }
1014
1015 #ifndef NO_PTHREADS
1016 static void *threaded_second_pass(void *data)
1017 {
1018         set_thread_data(data);
1019         for (;;) {
1020                 int i;
1021                 counter_lock();
1022                 display_progress(progress, nr_resolved_deltas);
1023                 counter_unlock();
1024                 work_lock();
1025                 while (nr_dispatched < nr_objects &&
1026                        is_delta_type(objects[nr_dispatched].type))
1027                         nr_dispatched++;
1028                 if (nr_dispatched >= nr_objects) {
1029                         work_unlock();
1030                         break;
1031                 }
1032                 i = nr_dispatched++;
1033                 work_unlock();
1034
1035                 resolve_base(&objects[i]);
1036         }
1037         return NULL;
1038 }
1039 #endif
1040
1041 /*
1042  * First pass:
1043  * - find locations of all objects;
1044  * - calculate SHA1 of all non-delta objects;
1045  * - remember base (SHA1 or offset) for all deltas.
1046  */
1047 static void parse_pack_objects(unsigned char *sha1)
1048 {
1049         int i, nr_delays = 0;
1050         struct delta_entry *delta = deltas;
1051         struct stat st;
1052
1053         if (verbose)
1054                 progress = start_progress(
1055                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1056                                 nr_objects);
1057         for (i = 0; i < nr_objects; i++) {
1058                 struct object_entry *obj = &objects[i];
1059                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1060                 obj->real_type = obj->type;
1061                 if (is_delta_type(obj->type)) {
1062                         nr_deltas++;
1063                         delta->obj_no = i;
1064                         delta++;
1065                 } else if (!data) {
1066                         /* large blobs, check later */
1067                         obj->real_type = OBJ_BAD;
1068                         nr_delays++;
1069                 } else
1070                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1071                 free(data);
1072                 display_progress(progress, i+1);
1073         }
1074         objects[i].idx.offset = consumed_bytes;
1075         stop_progress(&progress);
1076
1077         /* Check pack integrity */
1078         flush();
1079         git_SHA1_Final(sha1, &input_ctx);
1080         if (hashcmp(fill(20), sha1))
1081                 die(_("pack is corrupted (SHA1 mismatch)"));
1082         use(20);
1083
1084         /* If input_fd is a file, we should have reached its end now. */
1085         if (fstat(input_fd, &st))
1086                 die_errno(_("cannot fstat packfile"));
1087         if (S_ISREG(st.st_mode) &&
1088                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1089                 die(_("pack has junk at the end"));
1090
1091         for (i = 0; i < nr_objects; i++) {
1092                 struct object_entry *obj = &objects[i];
1093                 if (obj->real_type != OBJ_BAD)
1094                         continue;
1095                 obj->real_type = obj->type;
1096                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1097                 nr_delays--;
1098         }
1099         if (nr_delays)
1100                 die(_("confusion beyond insanity in parse_pack_objects()"));
1101 }
1102
1103 /*
1104  * Second pass:
1105  * - for all non-delta objects, look if it is used as a base for
1106  *   deltas;
1107  * - if used as a base, uncompress the object and apply all deltas,
1108  *   recursively checking if the resulting object is used as a base
1109  *   for some more deltas.
1110  */
1111 static void resolve_deltas(void)
1112 {
1113         int i;
1114
1115         if (!nr_deltas)
1116                 return;
1117
1118         /* Sort deltas by base SHA1/offset for fast searching */
1119         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1120               compare_delta_entry);
1121
1122         if (verbose)
1123                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1124
1125 #ifndef NO_PTHREADS
1126         nr_dispatched = 0;
1127         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1128                 init_thread();
1129                 for (i = 0; i < nr_threads; i++) {
1130                         int ret = pthread_create(&thread_data[i].thread, NULL,
1131                                                  threaded_second_pass, thread_data + i);
1132                         if (ret)
1133                                 die(_("unable to create thread: %s"),
1134                                     strerror(ret));
1135                 }
1136                 for (i = 0; i < nr_threads; i++)
1137                         pthread_join(thread_data[i].thread, NULL);
1138                 cleanup_thread();
1139                 return;
1140         }
1141 #endif
1142
1143         for (i = 0; i < nr_objects; i++) {
1144                 struct object_entry *obj = &objects[i];
1145
1146                 if (is_delta_type(obj->type))
1147                         continue;
1148                 resolve_base(obj);
1149                 display_progress(progress, nr_resolved_deltas);
1150         }
1151 }
1152
1153 /*
1154  * Third pass:
1155  * - append objects to convert thin pack to full pack if required
1156  * - write the final 20-byte SHA-1
1157  */
1158 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1159 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1160 {
1161         if (nr_deltas == nr_resolved_deltas) {
1162                 stop_progress(&progress);
1163                 /* Flush remaining pack final 20-byte SHA1. */
1164                 flush();
1165                 return;
1166         }
1167
1168         if (fix_thin_pack) {
1169                 struct sha1file *f;
1170                 unsigned char read_sha1[20], tail_sha1[20];
1171                 struct strbuf msg = STRBUF_INIT;
1172                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1173                 int nr_objects_initial = nr_objects;
1174                 if (nr_unresolved <= 0)
1175                         die(_("confusion beyond insanity"));
1176                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1177                 memset(objects + nr_objects + 1, 0,
1178                        nr_unresolved * sizeof(*objects));
1179                 f = sha1fd(output_fd, curr_pack);
1180                 fix_unresolved_deltas(f, nr_unresolved);
1181                 strbuf_addf(&msg, _("completed with %d local objects"),
1182                             nr_objects - nr_objects_initial);
1183                 stop_progress_msg(&progress, msg.buf);
1184                 strbuf_release(&msg);
1185                 sha1close(f, tail_sha1, 0);
1186                 hashcpy(read_sha1, pack_sha1);
1187                 fixup_pack_header_footer(output_fd, pack_sha1,
1188                                          curr_pack, nr_objects,
1189                                          read_sha1, consumed_bytes-20);
1190                 if (hashcmp(read_sha1, tail_sha1) != 0)
1191                         die(_("Unexpected tail checksum for %s "
1192                               "(disk corruption?)"), curr_pack);
1193         }
1194         if (nr_deltas != nr_resolved_deltas)
1195                 die(Q_("pack has %d unresolved delta",
1196                        "pack has %d unresolved deltas",
1197                        nr_deltas - nr_resolved_deltas),
1198                     nr_deltas - nr_resolved_deltas);
1199 }
1200
1201 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1202 {
1203         git_zstream stream;
1204         int status;
1205         unsigned char outbuf[4096];
1206
1207         memset(&stream, 0, sizeof(stream));
1208         git_deflate_init(&stream, zlib_compression_level);
1209         stream.next_in = in;
1210         stream.avail_in = size;
1211
1212         do {
1213                 stream.next_out = outbuf;
1214                 stream.avail_out = sizeof(outbuf);
1215                 status = git_deflate(&stream, Z_FINISH);
1216                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1217         } while (status == Z_OK);
1218
1219         if (status != Z_STREAM_END)
1220                 die(_("unable to deflate appended object (%d)"), status);
1221         size = stream.total_out;
1222         git_deflate_end(&stream);
1223         return size;
1224 }
1225
1226 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1227                                const unsigned char *sha1, void *buf,
1228                                unsigned long size, enum object_type type)
1229 {
1230         struct object_entry *obj = &objects[nr_objects++];
1231         unsigned char header[10];
1232         unsigned long s = size;
1233         int n = 0;
1234         unsigned char c = (type << 4) | (s & 15);
1235         s >>= 4;
1236         while (s) {
1237                 header[n++] = c | 0x80;
1238                 c = s & 0x7f;
1239                 s >>= 7;
1240         }
1241         header[n++] = c;
1242         crc32_begin(f);
1243         sha1write(f, header, n);
1244         obj[0].size = size;
1245         obj[0].hdr_size = n;
1246         obj[0].type = type;
1247         obj[0].real_type = type;
1248         obj[1].idx.offset = obj[0].idx.offset + n;
1249         obj[1].idx.offset += write_compressed(f, buf, size);
1250         obj[0].idx.crc32 = crc32_end(f);
1251         sha1flush(f);
1252         hashcpy(obj->idx.sha1, sha1);
1253         return obj;
1254 }
1255
1256 static int delta_pos_compare(const void *_a, const void *_b)
1257 {
1258         struct delta_entry *a = *(struct delta_entry **)_a;
1259         struct delta_entry *b = *(struct delta_entry **)_b;
1260         return a->obj_no - b->obj_no;
1261 }
1262
1263 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1264 {
1265         struct delta_entry **sorted_by_pos;
1266         int i, n = 0;
1267
1268         /*
1269          * Since many unresolved deltas may well be themselves base objects
1270          * for more unresolved deltas, we really want to include the
1271          * smallest number of base objects that would cover as much delta
1272          * as possible by picking the
1273          * trunc deltas first, allowing for other deltas to resolve without
1274          * additional base objects.  Since most base objects are to be found
1275          * before deltas depending on them, a good heuristic is to start
1276          * resolving deltas in the same order as their position in the pack.
1277          */
1278         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1279         for (i = 0; i < nr_deltas; i++) {
1280                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1281                         continue;
1282                 sorted_by_pos[n++] = &deltas[i];
1283         }
1284         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1285
1286         for (i = 0; i < n; i++) {
1287                 struct delta_entry *d = sorted_by_pos[i];
1288                 enum object_type type;
1289                 struct base_data *base_obj = alloc_base_data();
1290
1291                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1292                         continue;
1293                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1294                 if (!base_obj->data)
1295                         continue;
1296
1297                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1298                                 base_obj->size, typename(type)))
1299                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1300                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1301                                         base_obj->data, base_obj->size, type);
1302                 find_unresolved_deltas(base_obj);
1303                 display_progress(progress, nr_resolved_deltas);
1304         }
1305         free(sorted_by_pos);
1306 }
1307
1308 static void final(const char *final_pack_name, const char *curr_pack_name,
1309                   const char *final_index_name, const char *curr_index_name,
1310                   const char *keep_name, const char *keep_msg,
1311                   unsigned char *sha1)
1312 {
1313         const char *report = "pack";
1314         char name[PATH_MAX];
1315         int err;
1316
1317         if (!from_stdin) {
1318                 close(input_fd);
1319         } else {
1320                 fsync_or_die(output_fd, curr_pack_name);
1321                 err = close(output_fd);
1322                 if (err)
1323                         die_errno(_("error while closing pack file"));
1324         }
1325
1326         if (keep_msg) {
1327                 int keep_fd, keep_msg_len = strlen(keep_msg);
1328
1329                 if (!keep_name)
1330                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1331                 else
1332                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1333
1334                 if (keep_fd < 0) {
1335                         if (errno != EEXIST)
1336                                 die_errno(_("cannot write keep file '%s'"),
1337                                           keep_name ? keep_name : name);
1338                 } else {
1339                         if (keep_msg_len > 0) {
1340                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1341                                 write_or_die(keep_fd, "\n", 1);
1342                         }
1343                         if (close(keep_fd) != 0)
1344                                 die_errno(_("cannot close written keep file '%s'"),
1345                                           keep_name ? keep_name : name);
1346                         report = "keep";
1347                 }
1348         }
1349
1350         if (final_pack_name != curr_pack_name) {
1351                 if (!final_pack_name) {
1352                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1353                                  get_object_directory(), sha1_to_hex(sha1));
1354                         final_pack_name = name;
1355                 }
1356                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1357                         die(_("cannot store pack file"));
1358         } else if (from_stdin)
1359                 chmod(final_pack_name, 0444);
1360
1361         if (final_index_name != curr_index_name) {
1362                 if (!final_index_name) {
1363                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1364                                  get_object_directory(), sha1_to_hex(sha1));
1365                         final_index_name = name;
1366                 }
1367                 if (move_temp_to_file(curr_index_name, final_index_name))
1368                         die(_("cannot store index file"));
1369         } else
1370                 chmod(final_index_name, 0444);
1371
1372         if (!from_stdin) {
1373                 printf("%s\n", sha1_to_hex(sha1));
1374         } else {
1375                 char buf[48];
1376                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1377                                    report, sha1_to_hex(sha1));
1378                 write_or_die(1, buf, len);
1379
1380                 /*
1381                  * Let's just mimic git-unpack-objects here and write
1382                  * the last part of the input buffer to stdout.
1383                  */
1384                 while (input_len) {
1385                         err = xwrite(1, input_buffer + input_offset, input_len);
1386                         if (err <= 0)
1387                                 break;
1388                         input_len -= err;
1389                         input_offset += err;
1390                 }
1391         }
1392 }
1393
1394 static int git_index_pack_config(const char *k, const char *v, void *cb)
1395 {
1396         struct pack_idx_option *opts = cb;
1397
1398         if (!strcmp(k, "pack.indexversion")) {
1399                 opts->version = git_config_int(k, v);
1400                 if (opts->version > 2)
1401                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1402                 return 0;
1403         }
1404         if (!strcmp(k, "pack.threads")) {
1405                 nr_threads = git_config_int(k, v);
1406                 if (nr_threads < 0)
1407                         die(_("invalid number of threads specified (%d)"),
1408                             nr_threads);
1409 #ifdef NO_PTHREADS
1410                 if (nr_threads != 1)
1411                         warning(_("no threads support, ignoring %s"), k);
1412                 nr_threads = 1;
1413 #endif
1414                 return 0;
1415         }
1416         return git_default_config(k, v, cb);
1417 }
1418
1419 static int cmp_uint32(const void *a_, const void *b_)
1420 {
1421         uint32_t a = *((uint32_t *)a_);
1422         uint32_t b = *((uint32_t *)b_);
1423
1424         return (a < b) ? -1 : (a != b);
1425 }
1426
1427 static void read_v2_anomalous_offsets(struct packed_git *p,
1428                                       struct pack_idx_option *opts)
1429 {
1430         const uint32_t *idx1, *idx2;
1431         uint32_t i;
1432
1433         /* The address of the 4-byte offset table */
1434         idx1 = (((const uint32_t *)p->index_data)
1435                 + 2 /* 8-byte header */
1436                 + 256 /* fan out */
1437                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1438                 + p->num_objects /* CRC32 table */
1439                 );
1440
1441         /* The address of the 8-byte offset table */
1442         idx2 = idx1 + p->num_objects;
1443
1444         for (i = 0; i < p->num_objects; i++) {
1445                 uint32_t off = ntohl(idx1[i]);
1446                 if (!(off & 0x80000000))
1447                         continue;
1448                 off = off & 0x7fffffff;
1449                 if (idx2[off * 2])
1450                         continue;
1451                 /*
1452                  * The real offset is ntohl(idx2[off * 2]) in high 4
1453                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1454                  * octets.  But idx2[off * 2] is Zero!!!
1455                  */
1456                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1457                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1458         }
1459
1460         if (1 < opts->anomaly_nr)
1461                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1462 }
1463
1464 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1465 {
1466         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1467
1468         if (!p)
1469                 die(_("Cannot open existing pack file '%s'"), pack_name);
1470         if (open_pack_index(p))
1471                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1472
1473         /* Read the attributes from the existing idx file */
1474         opts->version = p->index_version;
1475
1476         if (opts->version == 2)
1477                 read_v2_anomalous_offsets(p, opts);
1478
1479         /*
1480          * Get rid of the idx file as we do not need it anymore.
1481          * NEEDSWORK: extract this bit from free_pack_by_name() in
1482          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1483          * know we haven't installed this pack (hence we never have
1484          * read anything from it).
1485          */
1486         close_pack_index(p);
1487         free(p);
1488 }
1489
1490 static void show_pack_info(int stat_only)
1491 {
1492         int i, baseobjects = nr_objects - nr_deltas;
1493         unsigned long *chain_histogram = NULL;
1494
1495         if (deepest_delta)
1496                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1497
1498         for (i = 0; i < nr_objects; i++) {
1499                 struct object_entry *obj = &objects[i];
1500
1501                 if (is_delta_type(obj->type))
1502                         chain_histogram[obj->delta_depth - 1]++;
1503                 if (stat_only)
1504                         continue;
1505                 printf("%s %-6s %lu %lu %"PRIuMAX,
1506                        sha1_to_hex(obj->idx.sha1),
1507                        typename(obj->real_type), obj->size,
1508                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1509                        (uintmax_t)obj->idx.offset);
1510                 if (is_delta_type(obj->type)) {
1511                         struct object_entry *bobj = &objects[obj->base_object_no];
1512                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1513                 }
1514                 putchar('\n');
1515         }
1516
1517         if (baseobjects)
1518                 printf_ln(Q_("non delta: %d object",
1519                              "non delta: %d objects",
1520                              baseobjects),
1521                           baseobjects);
1522         for (i = 0; i < deepest_delta; i++) {
1523                 if (!chain_histogram[i])
1524                         continue;
1525                 printf_ln(Q_("chain length = %d: %lu object",
1526                              "chain length = %d: %lu objects",
1527                              chain_histogram[i]),
1528                           i + 1,
1529                           chain_histogram[i]);
1530         }
1531 }
1532
1533 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1534 {
1535         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1536         const char *curr_index;
1537         const char *index_name = NULL, *pack_name = NULL;
1538         const char *keep_name = NULL, *keep_msg = NULL;
1539         struct strbuf index_name_buf = STRBUF_INIT,
1540                       keep_name_buf = STRBUF_INIT;
1541         struct pack_idx_entry **idx_objects;
1542         struct pack_idx_option opts;
1543         unsigned char pack_sha1[20];
1544         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1545
1546         if (argc == 2 && !strcmp(argv[1], "-h"))
1547                 usage(index_pack_usage);
1548
1549         check_replace_refs = 0;
1550
1551         reset_pack_idx_option(&opts);
1552         git_config(git_index_pack_config, &opts);
1553         if (prefix && chdir(prefix))
1554                 die(_("Cannot come back to cwd"));
1555
1556         for (i = 1; i < argc; i++) {
1557                 const char *arg = argv[i];
1558
1559                 if (*arg == '-') {
1560                         if (!strcmp(arg, "--stdin")) {
1561                                 from_stdin = 1;
1562                         } else if (!strcmp(arg, "--fix-thin")) {
1563                                 fix_thin_pack = 1;
1564                         } else if (!strcmp(arg, "--strict")) {
1565                                 strict = 1;
1566                                 do_fsck_object = 1;
1567                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1568                                 strict = 1;
1569                                 check_self_contained_and_connected = 1;
1570                         } else if (!strcmp(arg, "--verify")) {
1571                                 verify = 1;
1572                         } else if (!strcmp(arg, "--verify-stat")) {
1573                                 verify = 1;
1574                                 show_stat = 1;
1575                         } else if (!strcmp(arg, "--verify-stat-only")) {
1576                                 verify = 1;
1577                                 show_stat = 1;
1578                                 stat_only = 1;
1579                         } else if (!strcmp(arg, "--keep")) {
1580                                 keep_msg = "";
1581                         } else if (starts_with(arg, "--keep=")) {
1582                                 keep_msg = arg + 7;
1583                         } else if (starts_with(arg, "--threads=")) {
1584                                 char *end;
1585                                 nr_threads = strtoul(arg+10, &end, 0);
1586                                 if (!arg[10] || *end || nr_threads < 0)
1587                                         usage(index_pack_usage);
1588 #ifdef NO_PTHREADS
1589                                 if (nr_threads != 1)
1590                                         warning(_("no threads support, "
1591                                                   "ignoring %s"), arg);
1592                                 nr_threads = 1;
1593 #endif
1594                         } else if (starts_with(arg, "--pack_header=")) {
1595                                 struct pack_header *hdr;
1596                                 char *c;
1597
1598                                 hdr = (struct pack_header *)input_buffer;
1599                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1600                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1601                                 if (*c != ',')
1602                                         die(_("bad %s"), arg);
1603                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1604                                 if (*c)
1605                                         die(_("bad %s"), arg);
1606                                 input_len = sizeof(*hdr);
1607                         } else if (!strcmp(arg, "-v")) {
1608                                 verbose = 1;
1609                         } else if (!strcmp(arg, "-o")) {
1610                                 if (index_name || (i+1) >= argc)
1611                                         usage(index_pack_usage);
1612                                 index_name = argv[++i];
1613                         } else if (starts_with(arg, "--index-version=")) {
1614                                 char *c;
1615                                 opts.version = strtoul(arg + 16, &c, 10);
1616                                 if (opts.version > 2)
1617                                         die(_("bad %s"), arg);
1618                                 if (*c == ',')
1619                                         opts.off32_limit = strtoul(c+1, &c, 0);
1620                                 if (*c || opts.off32_limit & 0x80000000)
1621                                         die(_("bad %s"), arg);
1622                         } else
1623                                 usage(index_pack_usage);
1624                         continue;
1625                 }
1626
1627                 if (pack_name)
1628                         usage(index_pack_usage);
1629                 pack_name = arg;
1630         }
1631
1632         if (!pack_name && !from_stdin)
1633                 usage(index_pack_usage);
1634         if (fix_thin_pack && !from_stdin)
1635                 die(_("--fix-thin cannot be used without --stdin"));
1636         if (!index_name && pack_name) {
1637                 size_t len;
1638                 if (!strip_suffix(pack_name, ".pack", &len))
1639                         die(_("packfile name '%s' does not end with '.pack'"),
1640                             pack_name);
1641                 strbuf_add(&index_name_buf, pack_name, len);
1642                 strbuf_addstr(&index_name_buf, ".idx");
1643                 index_name = index_name_buf.buf;
1644         }
1645         if (keep_msg && !keep_name && pack_name) {
1646                 size_t len;
1647                 if (!strip_suffix(pack_name, ".pack", &len))
1648                         die(_("packfile name '%s' does not end with '.pack'"),
1649                             pack_name);
1650                 strbuf_add(&keep_name_buf, pack_name, len);
1651                 strbuf_addstr(&keep_name_buf, ".idx");
1652                 keep_name = keep_name_buf.buf;
1653         }
1654         if (verify) {
1655                 if (!index_name)
1656                         die(_("--verify with no packfile name given"));
1657                 read_idx_option(&opts, index_name);
1658                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1659         }
1660         if (strict)
1661                 opts.flags |= WRITE_IDX_STRICT;
1662
1663 #ifndef NO_PTHREADS
1664         if (!nr_threads) {
1665                 nr_threads = online_cpus();
1666                 /* An experiment showed that more threads does not mean faster */
1667                 if (nr_threads > 3)
1668                         nr_threads = 3;
1669         }
1670 #endif
1671
1672         curr_pack = open_pack_file(pack_name);
1673         parse_pack_header();
1674         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1675         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1676         parse_pack_objects(pack_sha1);
1677         resolve_deltas();
1678         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1679         free(deltas);
1680         if (strict)
1681                 foreign_nr = check_objects();
1682
1683         if (show_stat)
1684                 show_pack_info(stat_only);
1685
1686         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1687         for (i = 0; i < nr_objects; i++)
1688                 idx_objects[i] = &objects[i].idx;
1689         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1690         free(idx_objects);
1691
1692         if (!verify)
1693                 final(pack_name, curr_pack,
1694                       index_name, curr_index,
1695                       keep_name, keep_msg,
1696                       pack_sha1);
1697         else
1698                 close(input_fd);
1699         free(objects);
1700         strbuf_release(&index_name_buf);
1701         strbuf_release(&keep_name_buf);
1702         if (pack_name == NULL)
1703                 free((void *) curr_pack);
1704         if (index_name == NULL)
1705                 free((void *) curr_index);
1706
1707         /*
1708          * Let the caller know this pack is not self contained
1709          */
1710         if (check_self_contained_and_connected && foreign_nr)
1711                 return 1;
1712
1713         return 0;
1714 }