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 char hdr_size;
  22         signed char type;
  23         signed char real_type;
  24 };
  25
  26 struct object_stat {
  27         unsigned delta_depth;
  28         int base_object_no;
  29 };
  30
  31 struct base_data {
  32         struct base_data *base;
  33         struct base_data *child;
  34         struct object_entry *obj;
  35         void *data;
  36         unsigned long size;
  37         int ref_first, ref_last;
  38         int ofs_first, ofs_last;
  39 };
  40
  41 struct thread_local {
  42 #ifndef NO_PTHREADS
  43         pthread_t thread;
  44 #endif
  45         struct base_data *base_cache;
  46         size_t base_cache_used;
  47         int pack_fd;
  48 };
  49
  50 #define FLAG_LINK (1u<<20)
  51 #define FLAG_CHECKED (1u<<21)
  52
  53 struct ofs_delta_entry {
  54         off_t offset;
  55         int obj_no;
  56 };
  57
  58 struct ref_delta_entry {
  59         unsigned char sha1[20];
  60         int obj_no;
  61 };
  62
  63 static struct object_entry *objects;
  64 static struct object_stat *obj_stat;
  65 static struct ofs_delta_entry *ofs_deltas;
  66 static struct ref_delta_entry *ref_deltas;
  67 static struct thread_local nothread_data;
  68 static int nr_objects;
  69 static int nr_ofs_deltas;
  70 static int nr_ref_deltas;
  71 static int ref_deltas_alloc;
  72 static int nr_resolved_deltas;
  73 static int nr_threads;
  74
  75 static int from_stdin;
  76 static int strict;
  77 static int do_fsck_object;
  78 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  79 static int verbose;
  80 static int show_resolving_progress;
  81 static int show_stat;
  82 static int check_self_contained_and_connected;
  83
  84 static struct progress *progress;
  85
  86 /* We always read in 4kB chunks. */
  87 static unsigned char input_buffer[4096];
  88 static unsigned int input_offset, input_len;
  89 static off_t consumed_bytes;
  90 static off_t max_input_size;
  91 static unsigned deepest_delta;
  92 static git_SHA_CTX input_ctx;
  93 static uint32_t input_crc32;
  94 static int input_fd, output_fd;
  95 static const char *curr_pack;
  96
  97 #ifndef NO_PTHREADS
  98
  99 static struct thread_local *thread_data;
 100 static int nr_dispatched;
 101 static int threads_active;
 102
 103 static pthread_mutex_t read_mutex;
 104 #define read_lock()             lock_mutex(&read_mutex)
 105 #define read_unlock()           unlock_mutex(&read_mutex)
 106
 107 static pthread_mutex_t counter_mutex;
 108 #define counter_lock()          lock_mutex(&counter_mutex)
 109 #define counter_unlock()        unlock_mutex(&counter_mutex)
 110
 111 static pthread_mutex_t work_mutex;
 112 #define work_lock()             lock_mutex(&work_mutex)
 113 #define work_unlock()           unlock_mutex(&work_mutex)
 114
 115 static pthread_mutex_t deepest_delta_mutex;
 116 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 117 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 118
 119 static pthread_mutex_t type_cas_mutex;
 120 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 121 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 122
 123 static pthread_key_t key;
 124
 125 static inline void lock_mutex(pthread_mutex_t *mutex)
 126 {
 127         if (threads_active)
 128                 pthread_mutex_lock(mutex);
 129 }
 130
 131 static inline void unlock_mutex(pthread_mutex_t *mutex)
 132 {
 133         if (threads_active)
 134                 pthread_mutex_unlock(mutex);
 135 }
 136
 137 /*
 138  * Mutex and conditional variable can't be statically-initialized on Windows.
 139  */
 140 static void init_thread(void)
 141 {
 142         int i;
 143         init_recursive_mutex(&read_mutex);
 144         pthread_mutex_init(&counter_mutex, NULL);
 145         pthread_mutex_init(&work_mutex, NULL);
 146         pthread_mutex_init(&type_cas_mutex, NULL);
 147         if (show_stat)
 148                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 149         pthread_key_create(&key, NULL);
 150         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 151         for (i = 0; i < nr_threads; i++) {
 152                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 153                 if (thread_data[i].pack_fd == -1)
 154                         die_errno(_("unable to open %s"), curr_pack);
 155         }
 156
 157         threads_active = 1;
 158 }
 159
 160 static void cleanup_thread(void)
 161 {
 162         int i;
 163         if (!threads_active)
 164                 return;
 165         threads_active = 0;
 166         pthread_mutex_destroy(&read_mutex);
 167         pthread_mutex_destroy(&counter_mutex);
 168         pthread_mutex_destroy(&work_mutex);
 169         pthread_mutex_destroy(&type_cas_mutex);
 170         if (show_stat)
 171                 pthread_mutex_destroy(&deepest_delta_mutex);
 172         for (i = 0; i < nr_threads; i++)
 173                 close(thread_data[i].pack_fd);
 174         pthread_key_delete(key);
 175         free(thread_data);
 176 }
 177
 178 #else
 179
 180 #define read_lock()
 181 #define read_unlock()
 182
 183 #define counter_lock()
 184 #define counter_unlock()
 185
 186 #define work_lock()
 187 #define work_unlock()
 188
 189 #define deepest_delta_lock()
 190 #define deepest_delta_unlock()
 191
 192 #define type_cas_lock()
 193 #define type_cas_unlock()
 194
 195 #endif
 196
 197
 198 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 199 {
 200         if (!obj)
 201                 return -1;
 202
 203         if (type != OBJ_ANY && obj->type != type)
 204                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 205
 206         obj->flags |= FLAG_LINK;
 207         return 0;
 208 }
 209
 210 /* The content of each linked object must have been checked
 211    or it must be already present in the object database */
 212 static unsigned check_object(struct object *obj)
 213 {
 214         if (!obj)
 215                 return 0;
 216
 217         if (!(obj->flags & FLAG_LINK))
 218                 return 0;
 219
 220         if (!(obj->flags & FLAG_CHECKED)) {
 221                 unsigned long size;
 222                 int type = sha1_object_info(obj->oid.hash, &size);
 223                 if (type <= 0)
 224                         die(_("did not receive expected object %s"),
 225                               oid_to_hex(&obj->oid));
 226                 if (type != obj->type)
 227                         die(_("object %s: expected type %s, found %s"),
 228                             oid_to_hex(&obj->oid),
 229                             typename(obj->type), typename(type));
 230                 obj->flags |= FLAG_CHECKED;
 231                 return 1;
 232         }
 233
 234         return 0;
 235 }
 236
 237 static unsigned check_objects(void)
 238 {
 239         unsigned i, max, foreign_nr = 0;
 240
 241         max = get_max_object_index();
 242         for (i = 0; i < max; i++)
 243                 foreign_nr += check_object(get_indexed_object(i));
 244         return foreign_nr;
 245 }
 246
 247
 248 /* Discard current buffer used content. */
 249 static void flush(void)
 250 {
 251         if (input_offset) {
 252                 if (output_fd >= 0)
 253                         write_or_die(output_fd, input_buffer, input_offset);
 254                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 255                 memmove(input_buffer, input_buffer + input_offset, input_len);
 256                 input_offset = 0;
 257         }
 258 }
 259
 260 /*
 261  * Make sure at least "min" bytes are available in the buffer, and
 262  * return the pointer to the buffer.
 263  */
 264 static void *fill(int min)
 265 {
 266         if (min <= input_len)
 267                 return input_buffer + input_offset;
 268         if (min > sizeof(input_buffer))
 269                 die(Q_("cannot fill %d byte",
 270                        "cannot fill %d bytes",
 271                        min),
 272                     min);
 273         flush();
 274         do {
 275                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 276                                 sizeof(input_buffer) - input_len);
 277                 if (ret <= 0) {
 278                         if (!ret)
 279                                 die(_("early EOF"));
 280                         die_errno(_("read error on input"));
 281                 }
 282                 input_len += ret;
 283                 if (from_stdin)
 284                         display_throughput(progress, consumed_bytes + input_len);
 285         } while (input_len < min);
 286         return input_buffer;
 287 }
 288
 289 static void use(int bytes)
 290 {
 291         if (bytes > input_len)
 292                 die(_("used more bytes than were available"));
 293         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 294         input_len -= bytes;
 295         input_offset += bytes;
 296
 297         /* make sure off_t is sufficiently large not to wrap */
 298         if (signed_add_overflows(consumed_bytes, bytes))
 299                 die(_("pack too large for current definition of off_t"));
 300         consumed_bytes += bytes;
 301         if (max_input_size && consumed_bytes > max_input_size)
 302                 die(_("pack exceeds maximum allowed size"));
 303 }
 304
 305 static const char *open_pack_file(const char *pack_name)
 306 {
 307         if (from_stdin) {
 308                 input_fd = 0;
 309                 if (!pack_name) {
 310                         struct strbuf tmp_file = STRBUF_INIT;
 311                         output_fd = odb_mkstemp(&tmp_file,
 312                                                 "pack/tmp_pack_XXXXXX");
 313                         pack_name = strbuf_detach(&tmp_file, NULL);
 314                 } else {
 315                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 316                         if (output_fd < 0)
 317                                 die_errno(_("unable to create '%s'"), pack_name);
 318                 }
 319                 nothread_data.pack_fd = output_fd;
 320         } else {
 321                 input_fd = open(pack_name, O_RDONLY);
 322                 if (input_fd < 0)
 323                         die_errno(_("cannot open packfile '%s'"), pack_name);
 324                 output_fd = -1;
 325                 nothread_data.pack_fd = input_fd;
 326         }
 327         git_SHA1_Init(&input_ctx);
 328         return pack_name;
 329 }
 330
 331 static void parse_pack_header(void)
 332 {
 333         struct pack_header *hdr = fill(sizeof(struct pack_header));
 334
 335         /* Header consistency check */
 336         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 337                 die(_("pack signature mismatch"));
 338         if (!pack_version_ok(hdr->hdr_version))
 339                 die(_("pack version %"PRIu32" unsupported"),
 340                         ntohl(hdr->hdr_version));
 341
 342         nr_objects = ntohl(hdr->hdr_entries);
 343         use(sizeof(struct pack_header));
 344 }
 345
 346 static NORETURN void bad_object(off_t offset, const char *format,
 347                        ...) __attribute__((format (printf, 2, 3)));
 348
 349 static NORETURN void bad_object(off_t offset, const char *format, ...)
 350 {
 351         va_list params;
 352         char buf[1024];
 353
 354         va_start(params, format);
 355         vsnprintf(buf, sizeof(buf), format, params);
 356         va_end(params);
 357         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 358             (uintmax_t)offset, buf);
 359 }
 360
 361 static inline struct thread_local *get_thread_data(void)
 362 {
 363 #ifndef NO_PTHREADS
 364         if (threads_active)
 365                 return pthread_getspecific(key);
 366         assert(!threads_active &&
 367                "This should only be reached when all threads are gone");
 368 #endif
 369         return &nothread_data;
 370 }
 371
 372 #ifndef NO_PTHREADS
 373 static void set_thread_data(struct thread_local *data)
 374 {
 375         if (threads_active)
 376                 pthread_setspecific(key, data);
 377 }
 378 #endif
 379
 380 static struct base_data *alloc_base_data(void)
 381 {
 382         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 383         base->ref_last = -1;
 384         base->ofs_last = -1;
 385         return base;
 386 }
 387
 388 static void free_base_data(struct base_data *c)
 389 {
 390         if (c->data) {
 391                 FREE_AND_NULL(c->data);
 392                 get_thread_data()->base_cache_used -= c->size;
 393         }
 394 }
 395
 396 static void prune_base_data(struct base_data *retain)
 397 {
 398         struct base_data *b;
 399         struct thread_local *data = get_thread_data();
 400         for (b = data->base_cache;
 401              data->base_cache_used > delta_base_cache_limit && b;
 402              b = b->child) {
 403                 if (b->data && b != retain)
 404                         free_base_data(b);
 405         }
 406 }
 407
 408 static void link_base_data(struct base_data *base, struct base_data *c)
 409 {
 410         if (base)
 411                 base->child = c;
 412         else
 413                 get_thread_data()->base_cache = c;
 414
 415         c->base = base;
 416         c->child = NULL;
 417         if (c->data)
 418                 get_thread_data()->base_cache_used += c->size;
 419         prune_base_data(c);
 420 }
 421
 422 static void unlink_base_data(struct base_data *c)
 423 {
 424         struct base_data *base = c->base;
 425         if (base)
 426                 base->child = NULL;
 427         else
 428                 get_thread_data()->base_cache = NULL;
 429         free_base_data(c);
 430 }
 431
 432 static int is_delta_type(enum object_type type)
 433 {
 434         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 435 }
 436
 437 static void *unpack_entry_data(off_t offset, unsigned long size,
 438                                enum object_type type, unsigned char *sha1)
 439 {
 440         static char fixed_buf[8192];
 441         int status;
 442         git_zstream stream;
 443         void *buf;
 444         git_SHA_CTX c;
 445         char hdr[32];
 446         int hdrlen;
 447
 448         if (!is_delta_type(type)) {
 449                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(type), size) + 1;
 450                 git_SHA1_Init(&c);
 451                 git_SHA1_Update(&c, hdr, hdrlen);
 452         } else
 453                 sha1 = NULL;
 454         if (type == OBJ_BLOB && size > big_file_threshold)
 455                 buf = fixed_buf;
 456         else
 457                 buf = xmallocz(size);
 458
 459         memset(&stream, 0, sizeof(stream));
 460         git_inflate_init(&stream);
 461         stream.next_out = buf;
 462         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 463
 464         do {
 465                 unsigned char *last_out = stream.next_out;
 466                 stream.next_in = fill(1);
 467                 stream.avail_in = input_len;
 468                 status = git_inflate(&stream, 0);
 469                 use(input_len - stream.avail_in);
 470                 if (sha1)
 471                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 472                 if (buf == fixed_buf) {
 473                         stream.next_out = buf;
 474                         stream.avail_out = sizeof(fixed_buf);
 475                 }
 476         } while (status == Z_OK);
 477         if (stream.total_out != size || status != Z_STREAM_END)
 478                 bad_object(offset, _("inflate returned %d"), status);
 479         git_inflate_end(&stream);
 480         if (sha1)
 481                 git_SHA1_Final(sha1, &c);
 482         return buf == fixed_buf ? NULL : buf;
 483 }
 484
 485 static void *unpack_raw_entry(struct object_entry *obj,
 486                               off_t *ofs_offset,
 487                               unsigned char *ref_sha1,
 488                               unsigned char *sha1)
 489 {
 490         unsigned char *p;
 491         unsigned long size, c;
 492         off_t base_offset;
 493         unsigned shift;
 494         void *data;
 495
 496         obj->idx.offset = consumed_bytes;
 497         input_crc32 = crc32(0, NULL, 0);
 498
 499         p = fill(1);
 500         c = *p;
 501         use(1);
 502         obj->type = (c >> 4) & 7;
 503         size = (c & 15);
 504         shift = 4;
 505         while (c & 0x80) {
 506                 p = fill(1);
 507                 c = *p;
 508                 use(1);
 509                 size += (c & 0x7f) << shift;
 510                 shift += 7;
 511         }
 512         obj->size = size;
 513
 514         switch (obj->type) {
 515         case OBJ_REF_DELTA:
 516                 hashcpy(ref_sha1, fill(20));
 517                 use(20);
 518                 break;
 519         case OBJ_OFS_DELTA:
 520                 p = fill(1);
 521                 c = *p;
 522                 use(1);
 523                 base_offset = c & 127;
 524                 while (c & 128) {
 525                         base_offset += 1;
 526                         if (!base_offset || MSB(base_offset, 7))
 527                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 528                         p = fill(1);
 529                         c = *p;
 530                         use(1);
 531                         base_offset = (base_offset << 7) + (c & 127);
 532                 }
 533                 *ofs_offset = obj->idx.offset - base_offset;
 534                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 535                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 536                 break;
 537         case OBJ_COMMIT:
 538         case OBJ_TREE:
 539         case OBJ_BLOB:
 540         case OBJ_TAG:
 541                 break;
 542         default:
 543                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 544         }
 545         obj->hdr_size = consumed_bytes - obj->idx.offset;
 546
 547         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 548         obj->idx.crc32 = input_crc32;
 549         return data;
 550 }
 551
 552 static void *unpack_data(struct object_entry *obj,
 553                          int (*consume)(const unsigned char *, unsigned long, void *),
 554                          void *cb_data)
 555 {
 556         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 557         off_t len = obj[1].idx.offset - from;
 558         unsigned char *data, *inbuf;
 559         git_zstream stream;
 560         int status;
 561
 562         data = xmallocz(consume ? 64*1024 : obj->size);
 563         inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
 564
 565         memset(&stream, 0, sizeof(stream));
 566         git_inflate_init(&stream);
 567         stream.next_out = data;
 568         stream.avail_out = consume ? 64*1024 : obj->size;
 569
 570         do {
 571                 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
 572                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 573                 if (n < 0)
 574                         die_errno(_("cannot pread pack file"));
 575                 if (!n)
 576                         die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
 577                                "premature end of pack file, %"PRIuMAX" bytes missing",
 578                                (unsigned int)len),
 579                             (uintmax_t)len);
 580                 from += n;
 581                 len -= n;
 582                 stream.next_in = inbuf;
 583                 stream.avail_in = n;
 584                 if (!consume)
 585                         status = git_inflate(&stream, 0);
 586                 else {
 587                         do {
 588                                 status = git_inflate(&stream, 0);
 589                                 if (consume(data, stream.next_out - data, cb_data)) {
 590                                         free(inbuf);
 591                                         free(data);
 592                                         return NULL;
 593                                 }
 594                                 stream.next_out = data;
 595                                 stream.avail_out = 64*1024;
 596                         } while (status == Z_OK && stream.avail_in);
 597                 }
 598         } while (len && status == Z_OK && !stream.avail_in);
 599
 600         /* This has been inflated OK when first encountered, so... */
 601         if (status != Z_STREAM_END || stream.total_out != obj->size)
 602                 die(_("serious inflate inconsistency"));
 603
 604         git_inflate_end(&stream);
 605         free(inbuf);
 606         if (consume) {
 607                 FREE_AND_NULL(data);
 608         }
 609         return data;
 610 }
 611
 612 static void *get_data_from_pack(struct object_entry *obj)
 613 {
 614         return unpack_data(obj, NULL, NULL);
 615 }
 616
 617 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 618                                    enum object_type type1,
 619                                    enum object_type type2)
 620 {
 621         int cmp = type1 - type2;
 622         if (cmp)
 623                 return cmp;
 624         return offset1 < offset2 ? -1 :
 625                offset1 > offset2 ?  1 :
 626                0;
 627 }
 628
 629 static int find_ofs_delta(const off_t offset, enum object_type type)
 630 {
 631         int first = 0, last = nr_ofs_deltas;
 632
 633         while (first < last) {
 634                 int next = (first + last) / 2;
 635                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 636                 int cmp;
 637
 638                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 639                                               type, objects[delta->obj_no].type);
 640                 if (!cmp)
 641                         return next;
 642                 if (cmp < 0) {
 643                         last = next;
 644                         continue;
 645                 }
 646                 first = next+1;
 647         }
 648         return -first-1;
 649 }
 650
 651 static void find_ofs_delta_children(off_t offset,
 652                                     int *first_index, int *last_index,
 653                                     enum object_type type)
 654 {
 655         int first = find_ofs_delta(offset, type);
 656         int last = first;
 657         int end = nr_ofs_deltas - 1;
 658
 659         if (first < 0) {
 660                 *first_index = 0;
 661                 *last_index = -1;
 662                 return;
 663         }
 664         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 665                 --first;
 666         while (last < end && ofs_deltas[last + 1].offset == offset)
 667                 ++last;
 668         *first_index = first;
 669         *last_index = last;
 670 }
 671
 672 static int compare_ref_delta_bases(const unsigned char *sha1,
 673                                    const unsigned char *sha2,
 674                                    enum object_type type1,
 675                                    enum object_type type2)
 676 {
 677         int cmp = type1 - type2;
 678         if (cmp)
 679                 return cmp;
 680         return hashcmp(sha1, sha2);
 681 }
 682
 683 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
 684 {
 685         int first = 0, last = nr_ref_deltas;
 686
 687         while (first < last) {
 688                 int next = (first + last) / 2;
 689                 struct ref_delta_entry *delta = &ref_deltas[next];
 690                 int cmp;
 691
 692                 cmp = compare_ref_delta_bases(sha1, delta->sha1,
 693                                               type, objects[delta->obj_no].type);
 694                 if (!cmp)
 695                         return next;
 696                 if (cmp < 0) {
 697                         last = next;
 698                         continue;
 699                 }
 700                 first = next+1;
 701         }
 702         return -first-1;
 703 }
 704
 705 static void find_ref_delta_children(const unsigned char *sha1,
 706                                     int *first_index, int *last_index,
 707                                     enum object_type type)
 708 {
 709         int first = find_ref_delta(sha1, type);
 710         int last = first;
 711         int end = nr_ref_deltas - 1;
 712
 713         if (first < 0) {
 714                 *first_index = 0;
 715                 *last_index = -1;
 716                 return;
 717         }
 718         while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
 719                 --first;
 720         while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
 721                 ++last;
 722         *first_index = first;
 723         *last_index = last;
 724 }
 725
 726 struct compare_data {
 727         struct object_entry *entry;
 728         struct git_istream *st;
 729         unsigned char *buf;
 730         unsigned long buf_size;
 731 };
 732
 733 static int compare_objects(const unsigned char *buf, unsigned long size,
 734                            void *cb_data)
 735 {
 736         struct compare_data *data = cb_data;
 737
 738         if (data->buf_size < size) {
 739                 free(data->buf);
 740                 data->buf = xmalloc(size);
 741                 data->buf_size = size;
 742         }
 743
 744         while (size) {
 745                 ssize_t len = read_istream(data->st, data->buf, size);
 746                 if (len == 0)
 747                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 748                             oid_to_hex(&data->entry->idx.oid));
 749                 if (len < 0)
 750                         die(_("unable to read %s"),
 751                             oid_to_hex(&data->entry->idx.oid));
 752                 if (memcmp(buf, data->buf, len))
 753                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 754                             oid_to_hex(&data->entry->idx.oid));
 755                 size -= len;
 756                 buf += len;
 757         }
 758         return 0;
 759 }
 760
 761 static int check_collison(struct object_entry *entry)
 762 {
 763         struct compare_data data;
 764         enum object_type type;
 765         unsigned long size;
 766
 767         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 768                 return -1;
 769
 770         memset(&data, 0, sizeof(data));
 771         data.entry = entry;
 772         data.st = open_istream(entry->idx.oid.hash, &type, &size, NULL);
 773         if (!data.st)
 774                 return -1;
 775         if (size != entry->size || type != entry->type)
 776                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 777                     oid_to_hex(&entry->idx.oid));
 778         unpack_data(entry, compare_objects, &data);
 779         close_istream(data.st);
 780         free(data.buf);
 781         return 0;
 782 }
 783
 784 static void sha1_object(const void *data, struct object_entry *obj_entry,
 785                         unsigned long size, enum object_type type,
 786                         const struct object_id *oid)
 787 {
 788         void *new_data = NULL;
 789         int collision_test_needed = 0;
 790
 791         assert(data || obj_entry);
 792
 793         if (startup_info->have_repository) {
 794                 read_lock();
 795                 collision_test_needed = has_sha1_file_with_flags(oid->hash, HAS_SHA1_QUICK);
 796                 read_unlock();
 797         }
 798
 799         if (collision_test_needed && !data) {
 800                 read_lock();
 801                 if (!check_collison(obj_entry))
 802                         collision_test_needed = 0;
 803                 read_unlock();
 804         }
 805         if (collision_test_needed) {
 806                 void *has_data;
 807                 enum object_type has_type;
 808                 unsigned long has_size;
 809                 read_lock();
 810                 has_type = sha1_object_info(oid->hash, &has_size);
 811                 if (has_type < 0)
 812                         die(_("cannot read existing object info %s"), oid_to_hex(oid));
 813                 if (has_type != type || has_size != size)
 814                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 815                 has_data = read_sha1_file(oid->hash, &has_type, &has_size);
 816                 read_unlock();
 817                 if (!data)
 818                         data = new_data = get_data_from_pack(obj_entry);
 819                 if (!has_data)
 820                         die(_("cannot read existing object %s"), oid_to_hex(oid));
 821                 if (size != has_size || type != has_type ||
 822                     memcmp(data, has_data, size) != 0)
 823                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 824                 free(has_data);
 825         }
 826
 827         if (strict) {
 828                 read_lock();
 829                 if (type == OBJ_BLOB) {
 830                         struct blob *blob = lookup_blob(oid);
 831                         if (blob)
 832                                 blob->object.flags |= FLAG_CHECKED;
 833                         else
 834                                 die(_("invalid blob object %s"), oid_to_hex(oid));
 835                 } else {
 836                         struct object *obj;
 837                         int eaten;
 838                         void *buf = (void *) data;
 839
 840                         assert(data && "data can only be NULL for large _blobs_");
 841
 842                         /*
 843                          * we do not need to free the memory here, as the
 844                          * buf is deleted by the caller.
 845                          */
 846                         obj = parse_object_buffer(oid, type, size, buf,
 847                                                   &eaten);
 848                         if (!obj)
 849                                 die(_("invalid %s"), typename(type));
 850                         if (do_fsck_object &&
 851                             fsck_object(obj, buf, size, &fsck_options))
 852                                 die(_("Error in object"));
 853                         if (fsck_walk(obj, NULL, &fsck_options))
 854                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 855
 856                         if (obj->type == OBJ_TREE) {
 857                                 struct tree *item = (struct tree *) obj;
 858                                 item->buffer = NULL;
 859                                 obj->parsed = 0;
 860                         }
 861                         if (obj->type == OBJ_COMMIT) {
 862                                 struct commit *commit = (struct commit *) obj;
 863                                 if (detach_commit_buffer(commit, NULL) != data)
 864                                         die("BUG: parse_object_buffer transmogrified our buffer");
 865                         }
 866                         obj->flags |= FLAG_CHECKED;
 867                 }
 868                 read_unlock();
 869         }
 870
 871         free(new_data);
 872 }
 873
 874 /*
 875  * This function is part of find_unresolved_deltas(). There are two
 876  * walkers going in the opposite ways.
 877  *
 878  * The first one in find_unresolved_deltas() traverses down from
 879  * parent node to children, deflating nodes along the way. However,
 880  * memory for deflated nodes is limited by delta_base_cache_limit, so
 881  * at some point parent node's deflated content may be freed.
 882  *
 883  * The second walker is this function, which goes from current node up
 884  * to top parent if necessary to deflate the node. In normal
 885  * situation, its parent node would be already deflated, so it just
 886  * needs to apply delta.
 887  *
 888  * In the worst case scenario, parent node is no longer deflated because
 889  * we're running out of delta_base_cache_limit; we need to re-deflate
 890  * parents, possibly up to the top base.
 891  *
 892  * All deflated objects here are subject to be freed if we exceed
 893  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 894  * just need to make sure the last node is not freed.
 895  */
 896 static void *get_base_data(struct base_data *c)
 897 {
 898         if (!c->data) {
 899                 struct object_entry *obj = c->obj;
 900                 struct base_data **delta = NULL;
 901                 int delta_nr = 0, delta_alloc = 0;
 902
 903                 while (is_delta_type(c->obj->type) && !c->data) {
 904                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 905                         delta[delta_nr++] = c;
 906                         c = c->base;
 907                 }
 908                 if (!delta_nr) {
 909                         c->data = get_data_from_pack(obj);
 910                         c->size = obj->size;
 911                         get_thread_data()->base_cache_used += c->size;
 912                         prune_base_data(c);
 913                 }
 914                 for (; delta_nr > 0; delta_nr--) {
 915                         void *base, *raw;
 916                         c = delta[delta_nr - 1];
 917                         obj = c->obj;
 918                         base = get_base_data(c->base);
 919                         raw = get_data_from_pack(obj);
 920                         c->data = patch_delta(
 921                                 base, c->base->size,
 922                                 raw, obj->size,
 923                                 &c->size);
 924                         free(raw);
 925                         if (!c->data)
 926                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 927                         get_thread_data()->base_cache_used += c->size;
 928                         prune_base_data(c);
 929                 }
 930                 free(delta);
 931         }
 932         return c->data;
 933 }
 934
 935 static void resolve_delta(struct object_entry *delta_obj,
 936                           struct base_data *base, struct base_data *result)
 937 {
 938         void *base_data, *delta_data;
 939
 940         if (show_stat) {
 941                 int i = delta_obj - objects;
 942                 int j = base->obj - objects;
 943                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 944                 deepest_delta_lock();
 945                 if (deepest_delta < obj_stat[i].delta_depth)
 946                         deepest_delta = obj_stat[i].delta_depth;
 947                 deepest_delta_unlock();
 948                 obj_stat[i].base_object_no = j;
 949         }
 950         delta_data = get_data_from_pack(delta_obj);
 951         base_data = get_base_data(base);
 952         result->obj = delta_obj;
 953         result->data = patch_delta(base_data, base->size,
 954                                    delta_data, delta_obj->size, &result->size);
 955         free(delta_data);
 956         if (!result->data)
 957                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 958         hash_sha1_file(result->data, result->size,
 959                        typename(delta_obj->real_type),
 960                        delta_obj->idx.oid.hash);
 961         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 962                     &delta_obj->idx.oid);
 963         counter_lock();
 964         nr_resolved_deltas++;
 965         counter_unlock();
 966 }
 967
 968 /*
 969  * Standard boolean compare-and-swap: atomically check whether "*type" is
 970  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 971  * and return false.
 972  */
 973 static int compare_and_swap_type(signed char *type,
 974                                  enum object_type want,
 975                                  enum object_type set)
 976 {
 977         enum object_type old;
 978
 979         type_cas_lock();
 980         old = *type;
 981         if (old == want)
 982                 *type = set;
 983         type_cas_unlock();
 984
 985         return old == want;
 986 }
 987
 988 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 989                                                   struct base_data *prev_base)
 990 {
 991         if (base->ref_last == -1 && base->ofs_last == -1) {
 992                 find_ref_delta_children(base->obj->idx.oid.hash,
 993                                         &base->ref_first, &base->ref_last,
 994                                         OBJ_REF_DELTA);
 995
 996                 find_ofs_delta_children(base->obj->idx.offset,
 997                                         &base->ofs_first, &base->ofs_last,
 998                                         OBJ_OFS_DELTA);
 999
1000                 if (base->ref_last == -1 && base->ofs_last == -1) {
1001                         free(base->data);
1002                         return NULL;
1003                 }
1004
1005                 link_base_data(prev_base, base);
1006         }
1007
1008         if (base->ref_first <= base->ref_last) {
1009                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1010                 struct base_data *result = alloc_base_data();
1011
1012                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1013                                            base->obj->real_type))
1014                         die("BUG: child->real_type != OBJ_REF_DELTA");
1015
1016                 resolve_delta(child, base, result);
1017                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1018                         free_base_data(base);
1019
1020                 base->ref_first++;
1021                 return result;
1022         }
1023
1024         if (base->ofs_first <= base->ofs_last) {
1025                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1026                 struct base_data *result = alloc_base_data();
1027
1028                 assert(child->real_type == OBJ_OFS_DELTA);
1029                 child->real_type = base->obj->real_type;
1030                 resolve_delta(child, base, result);
1031                 if (base->ofs_first == base->ofs_last)
1032                         free_base_data(base);
1033
1034                 base->ofs_first++;
1035                 return result;
1036         }
1037
1038         unlink_base_data(base);
1039         return NULL;
1040 }
1041
1042 static void find_unresolved_deltas(struct base_data *base)
1043 {
1044         struct base_data *new_base, *prev_base = NULL;
1045         for (;;) {
1046                 new_base = find_unresolved_deltas_1(base, prev_base);
1047
1048                 if (new_base) {
1049                         prev_base = base;
1050                         base = new_base;
1051                 } else {
1052                         free(base);
1053                         base = prev_base;
1054                         if (!base)
1055                                 return;
1056                         prev_base = base->base;
1057                 }
1058         }
1059 }
1060
1061 static int compare_ofs_delta_entry(const void *a, const void *b)
1062 {
1063         const struct ofs_delta_entry *delta_a = a;
1064         const struct ofs_delta_entry *delta_b = b;
1065
1066         return delta_a->offset < delta_b->offset ? -1 :
1067                delta_a->offset > delta_b->offset ?  1 :
1068                0;
1069 }
1070
1071 static int compare_ref_delta_entry(const void *a, const void *b)
1072 {
1073         const struct ref_delta_entry *delta_a = a;
1074         const struct ref_delta_entry *delta_b = b;
1075
1076         return hashcmp(delta_a->sha1, delta_b->sha1);
1077 }
1078
1079 static void resolve_base(struct object_entry *obj)
1080 {
1081         struct base_data *base_obj = alloc_base_data();
1082         base_obj->obj = obj;
1083         base_obj->data = NULL;
1084         find_unresolved_deltas(base_obj);
1085 }
1086
1087 #ifndef NO_PTHREADS
1088 static void *threaded_second_pass(void *data)
1089 {
1090         set_thread_data(data);
1091         for (;;) {
1092                 int i;
1093                 counter_lock();
1094                 display_progress(progress, nr_resolved_deltas);
1095                 counter_unlock();
1096                 work_lock();
1097                 while (nr_dispatched < nr_objects &&
1098                        is_delta_type(objects[nr_dispatched].type))
1099                         nr_dispatched++;
1100                 if (nr_dispatched >= nr_objects) {
1101                         work_unlock();
1102                         break;
1103                 }
1104                 i = nr_dispatched++;
1105                 work_unlock();
1106
1107                 resolve_base(&objects[i]);
1108         }
1109         return NULL;
1110 }
1111 #endif
1112
1113 /*
1114  * First pass:
1115  * - find locations of all objects;
1116  * - calculate SHA1 of all non-delta objects;
1117  * - remember base (SHA1 or offset) for all deltas.
1118  */
1119 static void parse_pack_objects(unsigned char *sha1)
1120 {
1121         int i, nr_delays = 0;
1122         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1123         unsigned char ref_delta_sha1[20];
1124         struct stat st;
1125
1126         if (verbose)
1127                 progress = start_progress(
1128                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1129                                 nr_objects);
1130         for (i = 0; i < nr_objects; i++) {
1131                 struct object_entry *obj = &objects[i];
1132                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1133                                               ref_delta_sha1,
1134                                               obj->idx.oid.hash);
1135                 obj->real_type = obj->type;
1136                 if (obj->type == OBJ_OFS_DELTA) {
1137                         nr_ofs_deltas++;
1138                         ofs_delta->obj_no = i;
1139                         ofs_delta++;
1140                 } else if (obj->type == OBJ_REF_DELTA) {
1141                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1142                         hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1);
1143                         ref_deltas[nr_ref_deltas].obj_no = i;
1144                         nr_ref_deltas++;
1145                 } else if (!data) {
1146                         /* large blobs, check later */
1147                         obj->real_type = OBJ_BAD;
1148                         nr_delays++;
1149                 } else
1150                         sha1_object(data, NULL, obj->size, obj->type,
1151                                     &obj->idx.oid);
1152                 free(data);
1153                 display_progress(progress, i+1);
1154         }
1155         objects[i].idx.offset = consumed_bytes;
1156         stop_progress(&progress);
1157
1158         /* Check pack integrity */
1159         flush();
1160         git_SHA1_Final(sha1, &input_ctx);
1161         if (hashcmp(fill(20), sha1))
1162                 die(_("pack is corrupted (SHA1 mismatch)"));
1163         use(20);
1164
1165         /* If input_fd is a file, we should have reached its end now. */
1166         if (fstat(input_fd, &st))
1167                 die_errno(_("cannot fstat packfile"));
1168         if (S_ISREG(st.st_mode) &&
1169                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1170                 die(_("pack has junk at the end"));
1171
1172         for (i = 0; i < nr_objects; i++) {
1173                 struct object_entry *obj = &objects[i];
1174                 if (obj->real_type != OBJ_BAD)
1175                         continue;
1176                 obj->real_type = obj->type;
1177                 sha1_object(NULL, obj, obj->size, obj->type,
1178                             &obj->idx.oid);
1179                 nr_delays--;
1180         }
1181         if (nr_delays)
1182                 die(_("confusion beyond insanity in parse_pack_objects()"));
1183 }
1184
1185 /*
1186  * Second pass:
1187  * - for all non-delta objects, look if it is used as a base for
1188  *   deltas;
1189  * - if used as a base, uncompress the object and apply all deltas,
1190  *   recursively checking if the resulting object is used as a base
1191  *   for some more deltas.
1192  */
1193 static void resolve_deltas(void)
1194 {
1195         int i;
1196
1197         if (!nr_ofs_deltas && !nr_ref_deltas)
1198                 return;
1199
1200         /* Sort deltas by base SHA1/offset for fast searching */
1201         QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1202         QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1203
1204         if (verbose || show_resolving_progress)
1205                 progress = start_progress(_("Resolving deltas"),
1206                                           nr_ref_deltas + nr_ofs_deltas);
1207
1208 #ifndef NO_PTHREADS
1209         nr_dispatched = 0;
1210         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1211                 init_thread();
1212                 for (i = 0; i < nr_threads; i++) {
1213                         int ret = pthread_create(&thread_data[i].thread, NULL,
1214                                                  threaded_second_pass, thread_data + i);
1215                         if (ret)
1216                                 die(_("unable to create thread: %s"),
1217                                     strerror(ret));
1218                 }
1219                 for (i = 0; i < nr_threads; i++)
1220                         pthread_join(thread_data[i].thread, NULL);
1221                 cleanup_thread();
1222                 return;
1223         }
1224 #endif
1225
1226         for (i = 0; i < nr_objects; i++) {
1227                 struct object_entry *obj = &objects[i];
1228
1229                 if (is_delta_type(obj->type))
1230                         continue;
1231                 resolve_base(obj);
1232                 display_progress(progress, nr_resolved_deltas);
1233         }
1234 }
1235
1236 /*
1237  * Third pass:
1238  * - append objects to convert thin pack to full pack if required
1239  * - write the final 20-byte SHA-1
1240  */
1241 static void fix_unresolved_deltas(struct sha1file *f);
1242 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1243 {
1244         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1245                 stop_progress(&progress);
1246                 /* Flush remaining pack final 20-byte SHA1. */
1247                 flush();
1248                 return;
1249         }
1250
1251         if (fix_thin_pack) {
1252                 struct sha1file *f;
1253                 unsigned char read_sha1[20], tail_sha1[20];
1254                 struct strbuf msg = STRBUF_INIT;
1255                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1256                 int nr_objects_initial = nr_objects;
1257                 if (nr_unresolved <= 0)
1258                         die(_("confusion beyond insanity"));
1259                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1260                 memset(objects + nr_objects + 1, 0,
1261                        nr_unresolved * sizeof(*objects));
1262                 f = sha1fd(output_fd, curr_pack);
1263                 fix_unresolved_deltas(f);
1264                 strbuf_addf(&msg, Q_("completed with %d local object",
1265                                      "completed with %d local objects",
1266                                      nr_objects - nr_objects_initial),
1267                             nr_objects - nr_objects_initial);
1268                 stop_progress_msg(&progress, msg.buf);
1269                 strbuf_release(&msg);
1270                 sha1close(f, tail_sha1, 0);
1271                 hashcpy(read_sha1, pack_sha1);
1272                 fixup_pack_header_footer(output_fd, pack_sha1,
1273                                          curr_pack, nr_objects,
1274                                          read_sha1, consumed_bytes-20);
1275                 if (hashcmp(read_sha1, tail_sha1) != 0)
1276                         die(_("Unexpected tail checksum for %s "
1277                               "(disk corruption?)"), curr_pack);
1278         }
1279         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1280                 die(Q_("pack has %d unresolved delta",
1281                        "pack has %d unresolved deltas",
1282                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1283                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1284 }
1285
1286 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1287 {
1288         git_zstream stream;
1289         int status;
1290         unsigned char outbuf[4096];
1291
1292         git_deflate_init(&stream, zlib_compression_level);
1293         stream.next_in = in;
1294         stream.avail_in = size;
1295
1296         do {
1297                 stream.next_out = outbuf;
1298                 stream.avail_out = sizeof(outbuf);
1299                 status = git_deflate(&stream, Z_FINISH);
1300                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1301         } while (status == Z_OK);
1302
1303         if (status != Z_STREAM_END)
1304                 die(_("unable to deflate appended object (%d)"), status);
1305         size = stream.total_out;
1306         git_deflate_end(&stream);
1307         return size;
1308 }
1309
1310 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1311                                const unsigned char *sha1, void *buf,
1312                                unsigned long size, enum object_type type)
1313 {
1314         struct object_entry *obj = &objects[nr_objects++];
1315         unsigned char header[10];
1316         unsigned long s = size;
1317         int n = 0;
1318         unsigned char c = (type << 4) | (s & 15);
1319         s >>= 4;
1320         while (s) {
1321                 header[n++] = c | 0x80;
1322                 c = s & 0x7f;
1323                 s >>= 7;
1324         }
1325         header[n++] = c;
1326         crc32_begin(f);
1327         sha1write(f, header, n);
1328         obj[0].size = size;
1329         obj[0].hdr_size = n;
1330         obj[0].type = type;
1331         obj[0].real_type = type;
1332         obj[1].idx.offset = obj[0].idx.offset + n;
1333         obj[1].idx.offset += write_compressed(f, buf, size);
1334         obj[0].idx.crc32 = crc32_end(f);
1335         sha1flush(f);
1336         hashcpy(obj->idx.oid.hash, sha1);
1337         return obj;
1338 }
1339
1340 static int delta_pos_compare(const void *_a, const void *_b)
1341 {
1342         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1343         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1344         return a->obj_no - b->obj_no;
1345 }
1346
1347 static void fix_unresolved_deltas(struct sha1file *f)
1348 {
1349         struct ref_delta_entry **sorted_by_pos;
1350         int i;
1351
1352         /*
1353          * Since many unresolved deltas may well be themselves base objects
1354          * for more unresolved deltas, we really want to include the
1355          * smallest number of base objects that would cover as much delta
1356          * as possible by picking the
1357          * trunc deltas first, allowing for other deltas to resolve without
1358          * additional base objects.  Since most base objects are to be found
1359          * before deltas depending on them, a good heuristic is to start
1360          * resolving deltas in the same order as their position in the pack.
1361          */
1362         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1363         for (i = 0; i < nr_ref_deltas; i++)
1364                 sorted_by_pos[i] = &ref_deltas[i];
1365         QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
1366
1367         for (i = 0; i < nr_ref_deltas; i++) {
1368                 struct ref_delta_entry *d = sorted_by_pos[i];
1369                 enum object_type type;
1370                 struct base_data *base_obj = alloc_base_data();
1371
1372                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1373                         continue;
1374                 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1375                 if (!base_obj->data)
1376                         continue;
1377
1378                 if (check_sha1_signature(d->sha1, base_obj->data,
1379                                 base_obj->size, typename(type)))
1380                         die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1381                 base_obj->obj = append_obj_to_pack(f, d->sha1,
1382                                         base_obj->data, base_obj->size, type);
1383                 find_unresolved_deltas(base_obj);
1384                 display_progress(progress, nr_resolved_deltas);
1385         }
1386         free(sorted_by_pos);
1387 }
1388
1389 static void final(const char *final_pack_name, const char *curr_pack_name,
1390                   const char *final_index_name, const char *curr_index_name,
1391                   const char *keep_name, const char *keep_msg,
1392                   unsigned char *sha1)
1393 {
1394         const char *report = "pack";
1395         struct strbuf pack_name = STRBUF_INIT;
1396         struct strbuf index_name = STRBUF_INIT;
1397         struct strbuf keep_name_buf = STRBUF_INIT;
1398         int err;
1399
1400         if (!from_stdin) {
1401                 close(input_fd);
1402         } else {
1403                 fsync_or_die(output_fd, curr_pack_name);
1404                 err = close(output_fd);
1405                 if (err)
1406                         die_errno(_("error while closing pack file"));
1407         }
1408
1409         if (keep_msg) {
1410                 int keep_fd, keep_msg_len = strlen(keep_msg);
1411
1412                 if (!keep_name)
1413                         keep_name = odb_pack_name(&keep_name_buf, sha1, "keep");
1414
1415                 keep_fd = odb_pack_keep(keep_name);
1416                 if (keep_fd < 0) {
1417                         if (errno != EEXIST)
1418                                 die_errno(_("cannot write keep file '%s'"),
1419                                           keep_name);
1420                 } else {
1421                         if (keep_msg_len > 0) {
1422                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1423                                 write_or_die(keep_fd, "\n", 1);
1424                         }
1425                         if (close(keep_fd) != 0)
1426                                 die_errno(_("cannot close written keep file '%s'"),
1427                                           keep_name);
1428                         report = "keep";
1429                 }
1430         }
1431
1432         if (final_pack_name != curr_pack_name) {
1433                 if (!final_pack_name)
1434                         final_pack_name = odb_pack_name(&pack_name, sha1, "pack");
1435                 if (finalize_object_file(curr_pack_name, final_pack_name))
1436                         die(_("cannot store pack file"));
1437         } else if (from_stdin)
1438                 chmod(final_pack_name, 0444);
1439
1440         if (final_index_name != curr_index_name) {
1441                 if (!final_index_name)
1442                         final_index_name = odb_pack_name(&index_name, sha1, "idx");
1443                 if (finalize_object_file(curr_index_name, final_index_name))
1444                         die(_("cannot store index file"));
1445         } else
1446                 chmod(final_index_name, 0444);
1447
1448         if (!from_stdin) {
1449                 printf("%s\n", sha1_to_hex(sha1));
1450         } else {
1451                 struct strbuf buf = STRBUF_INIT;
1452
1453                 strbuf_addf(&buf, "%s\t%s\n", report, sha1_to_hex(sha1));
1454                 write_or_die(1, buf.buf, buf.len);
1455                 strbuf_release(&buf);
1456
1457                 /*
1458                  * Let's just mimic git-unpack-objects here and write
1459                  * the last part of the input buffer to stdout.
1460                  */
1461                 while (input_len) {
1462                         err = xwrite(1, input_buffer + input_offset, input_len);
1463                         if (err <= 0)
1464                                 break;
1465                         input_len -= err;
1466                         input_offset += err;
1467                 }
1468         }
1469
1470         strbuf_release(&index_name);
1471         strbuf_release(&pack_name);
1472         strbuf_release(&keep_name_buf);
1473 }
1474
1475 static int git_index_pack_config(const char *k, const char *v, void *cb)
1476 {
1477         struct pack_idx_option *opts = cb;
1478
1479         if (!strcmp(k, "pack.indexversion")) {
1480                 opts->version = git_config_int(k, v);
1481                 if (opts->version > 2)
1482                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1483                 return 0;
1484         }
1485         if (!strcmp(k, "pack.threads")) {
1486                 nr_threads = git_config_int(k, v);
1487                 if (nr_threads < 0)
1488                         die(_("invalid number of threads specified (%d)"),
1489                             nr_threads);
1490 #ifdef NO_PTHREADS
1491                 if (nr_threads != 1)
1492                         warning(_("no threads support, ignoring %s"), k);
1493                 nr_threads = 1;
1494 #endif
1495                 return 0;
1496         }
1497         return git_default_config(k, v, cb);
1498 }
1499
1500 static int cmp_uint32(const void *a_, const void *b_)
1501 {
1502         uint32_t a = *((uint32_t *)a_);
1503         uint32_t b = *((uint32_t *)b_);
1504
1505         return (a < b) ? -1 : (a != b);
1506 }
1507
1508 static void read_v2_anomalous_offsets(struct packed_git *p,
1509                                       struct pack_idx_option *opts)
1510 {
1511         const uint32_t *idx1, *idx2;
1512         uint32_t i;
1513
1514         /* The address of the 4-byte offset table */
1515         idx1 = (((const uint32_t *)p->index_data)
1516                 + 2 /* 8-byte header */
1517                 + 256 /* fan out */
1518                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1519                 + p->num_objects /* CRC32 table */
1520                 );
1521
1522         /* The address of the 8-byte offset table */
1523         idx2 = idx1 + p->num_objects;
1524
1525         for (i = 0; i < p->num_objects; i++) {
1526                 uint32_t off = ntohl(idx1[i]);
1527                 if (!(off & 0x80000000))
1528                         continue;
1529                 off = off & 0x7fffffff;
1530                 check_pack_index_ptr(p, &idx2[off * 2]);
1531                 if (idx2[off * 2])
1532                         continue;
1533                 /*
1534                  * The real offset is ntohl(idx2[off * 2]) in high 4
1535                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1536                  * octets.  But idx2[off * 2] is Zero!!!
1537                  */
1538                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1539                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1540         }
1541
1542         QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32);
1543 }
1544
1545 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1546 {
1547         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1548
1549         if (!p)
1550                 die(_("Cannot open existing pack file '%s'"), pack_name);
1551         if (open_pack_index(p))
1552                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1553
1554         /* Read the attributes from the existing idx file */
1555         opts->version = p->index_version;
1556
1557         if (opts->version == 2)
1558                 read_v2_anomalous_offsets(p, opts);
1559
1560         /*
1561          * Get rid of the idx file as we do not need it anymore.
1562          * NEEDSWORK: extract this bit from free_pack_by_name() in
1563          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1564          * know we haven't installed this pack (hence we never have
1565          * read anything from it).
1566          */
1567         close_pack_index(p);
1568         free(p);
1569 }
1570
1571 static void show_pack_info(int stat_only)
1572 {
1573         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1574         unsigned long *chain_histogram = NULL;
1575
1576         if (deepest_delta)
1577                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1578
1579         for (i = 0; i < nr_objects; i++) {
1580                 struct object_entry *obj = &objects[i];
1581
1582                 if (is_delta_type(obj->type))
1583                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1584                 if (stat_only)
1585                         continue;
1586                 printf("%s %-6s %lu %lu %"PRIuMAX,
1587                        oid_to_hex(&obj->idx.oid),
1588                        typename(obj->real_type), obj->size,
1589                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1590                        (uintmax_t)obj->idx.offset);
1591                 if (is_delta_type(obj->type)) {
1592                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1593                         printf(" %u %s", obj_stat[i].delta_depth,
1594                                oid_to_hex(&bobj->idx.oid));
1595                 }
1596                 putchar('\n');
1597         }
1598
1599         if (baseobjects)
1600                 printf_ln(Q_("non delta: %d object",
1601                              "non delta: %d objects",
1602                              baseobjects),
1603                           baseobjects);
1604         for (i = 0; i < deepest_delta; i++) {
1605                 if (!chain_histogram[i])
1606                         continue;
1607                 printf_ln(Q_("chain length = %d: %lu object",
1608                              "chain length = %d: %lu objects",
1609                              chain_histogram[i]),
1610                           i + 1,
1611                           chain_histogram[i]);
1612         }
1613 }
1614
1615 static const char *derive_filename(const char *pack_name, const char *suffix,
1616                                    struct strbuf *buf)
1617 {
1618         size_t len;
1619         if (!strip_suffix(pack_name, ".pack", &len))
1620                 die(_("packfile name '%s' does not end with '.pack'"),
1621                     pack_name);
1622         strbuf_add(buf, pack_name, len);
1623         strbuf_addstr(buf, suffix);
1624         return buf->buf;
1625 }
1626
1627 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1628 {
1629         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1630         const char *curr_index;
1631         const char *index_name = NULL, *pack_name = NULL;
1632         const char *keep_name = NULL, *keep_msg = NULL;
1633         struct strbuf index_name_buf = STRBUF_INIT,
1634                       keep_name_buf = STRBUF_INIT;
1635         struct pack_idx_entry **idx_objects;
1636         struct pack_idx_option opts;
1637         unsigned char pack_sha1[20];
1638         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1639         int report_end_of_input = 0;
1640
1641         if (argc == 2 && !strcmp(argv[1], "-h"))
1642                 usage(index_pack_usage);
1643
1644         check_replace_refs = 0;
1645         fsck_options.walk = mark_link;
1646
1647         reset_pack_idx_option(&opts);
1648         git_config(git_index_pack_config, &opts);
1649         if (prefix && chdir(prefix))
1650                 die(_("Cannot come back to cwd"));
1651
1652         for (i = 1; i < argc; i++) {
1653                 const char *arg = argv[i];
1654
1655                 if (*arg == '-') {
1656                         if (!strcmp(arg, "--stdin")) {
1657                                 from_stdin = 1;
1658                         } else if (!strcmp(arg, "--fix-thin")) {
1659                                 fix_thin_pack = 1;
1660                         } else if (!strcmp(arg, "--strict")) {
1661                                 strict = 1;
1662                                 do_fsck_object = 1;
1663                         } else if (skip_prefix(arg, "--strict=", &arg)) {
1664                                 strict = 1;
1665                                 do_fsck_object = 1;
1666                                 fsck_set_msg_types(&fsck_options, arg);
1667                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1668                                 strict = 1;
1669                                 check_self_contained_and_connected = 1;
1670                         } else if (!strcmp(arg, "--verify")) {
1671                                 verify = 1;
1672                         } else if (!strcmp(arg, "--verify-stat")) {
1673                                 verify = 1;
1674                                 show_stat = 1;
1675                         } else if (!strcmp(arg, "--verify-stat-only")) {
1676                                 verify = 1;
1677                                 show_stat = 1;
1678                                 stat_only = 1;
1679                         } else if (!strcmp(arg, "--keep")) {
1680                                 keep_msg = "";
1681                         } else if (starts_with(arg, "--keep=")) {
1682                                 keep_msg = arg + 7;
1683                         } else if (starts_with(arg, "--threads=")) {
1684                                 char *end;
1685                                 nr_threads = strtoul(arg+10, &end, 0);
1686                                 if (!arg[10] || *end || nr_threads < 0)
1687                                         usage(index_pack_usage);
1688 #ifdef NO_PTHREADS
1689                                 if (nr_threads != 1)
1690                                         warning(_("no threads support, "
1691                                                   "ignoring %s"), arg);
1692                                 nr_threads = 1;
1693 #endif
1694                         } else if (starts_with(arg, "--pack_header=")) {
1695                                 struct pack_header *hdr;
1696                                 char *c;
1697
1698                                 hdr = (struct pack_header *)input_buffer;
1699                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1700                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1701                                 if (*c != ',')
1702                                         die(_("bad %s"), arg);
1703                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1704                                 if (*c)
1705                                         die(_("bad %s"), arg);
1706                                 input_len = sizeof(*hdr);
1707                         } else if (!strcmp(arg, "-v")) {
1708                                 verbose = 1;
1709                         } else if (!strcmp(arg, "--show-resolving-progress")) {
1710                                 show_resolving_progress = 1;
1711                         } else if (!strcmp(arg, "--report-end-of-input")) {
1712                                 report_end_of_input = 1;
1713                         } else if (!strcmp(arg, "-o")) {
1714                                 if (index_name || (i+1) >= argc)
1715                                         usage(index_pack_usage);
1716                                 index_name = argv[++i];
1717                         } else if (starts_with(arg, "--index-version=")) {
1718                                 char *c;
1719                                 opts.version = strtoul(arg + 16, &c, 10);
1720                                 if (opts.version > 2)
1721                                         die(_("bad %s"), arg);
1722                                 if (*c == ',')
1723                                         opts.off32_limit = strtoul(c+1, &c, 0);
1724                                 if (*c || opts.off32_limit & 0x80000000)
1725                                         die(_("bad %s"), arg);
1726                         } else if (skip_prefix(arg, "--max-input-size=", &arg)) {
1727                                 max_input_size = strtoumax(arg, NULL, 10);
1728                         } else
1729                                 usage(index_pack_usage);
1730                         continue;
1731                 }
1732
1733                 if (pack_name)
1734                         usage(index_pack_usage);
1735                 pack_name = arg;
1736         }
1737
1738         if (!pack_name && !from_stdin)
1739                 usage(index_pack_usage);
1740         if (fix_thin_pack && !from_stdin)
1741                 die(_("--fix-thin cannot be used without --stdin"));
1742         if (from_stdin && !startup_info->have_repository)
1743                 die(_("--stdin requires a git repository"));
1744         if (!index_name && pack_name)
1745                 index_name = derive_filename(pack_name, ".idx", &index_name_buf);
1746         if (keep_msg && !keep_name && pack_name)
1747                 keep_name = derive_filename(pack_name, ".keep", &keep_name_buf);
1748
1749         if (verify) {
1750                 if (!index_name)
1751                         die(_("--verify with no packfile name given"));
1752                 read_idx_option(&opts, index_name);
1753                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1754         }
1755         if (strict)
1756                 opts.flags |= WRITE_IDX_STRICT;
1757
1758 #ifndef NO_PTHREADS
1759         if (!nr_threads) {
1760                 nr_threads = online_cpus();
1761                 /* An experiment showed that more threads does not mean faster */
1762                 if (nr_threads > 3)
1763                         nr_threads = 3;
1764         }
1765 #endif
1766
1767         curr_pack = open_pack_file(pack_name);
1768         parse_pack_header();
1769         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1770         if (show_stat)
1771                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1772         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1773         parse_pack_objects(pack_sha1);
1774         if (report_end_of_input)
1775                 write_in_full(2, "\0", 1);
1776         resolve_deltas();
1777         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1778         free(ofs_deltas);
1779         free(ref_deltas);
1780         if (strict)
1781                 foreign_nr = check_objects();
1782
1783         if (show_stat)
1784                 show_pack_info(stat_only);
1785
1786         ALLOC_ARRAY(idx_objects, nr_objects);
1787         for (i = 0; i < nr_objects; i++)
1788                 idx_objects[i] = &objects[i].idx;
1789         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1790         free(idx_objects);
1791
1792         if (!verify)
1793                 final(pack_name, curr_pack,
1794                       index_name, curr_index,
1795                       keep_name, keep_msg,
1796                       pack_sha1);
1797         else
1798                 close(input_fd);
1799         free(objects);
1800         strbuf_release(&index_name_buf);
1801         strbuf_release(&keep_name_buf);
1802         if (pack_name == NULL)
1803                 free((void *) curr_pack);
1804         if (index_name == NULL)
1805                 free((void *) curr_index);
1806
1807         /*
1808          * Let the caller know this pack is not self contained
1809          */
1810         if (check_self_contained_and_connected && foreign_nr)
1811                 return 1;
1812
1813         return 0;
1814 }