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

   1 #include "builtin.h"
   2 #include "config.h"
   3 #include "delta.h"
   4 #include "pack.h"
   5 #include "csum-file.h"
   6 #include "blob.h"
   7 #include "commit.h"
   8 #include "tag.h"
   9 #include "tree.h"
  10 #include "progress.h"
  11 #include "fsck.h"
  12 #include "exec_cmd.h"
  13 #include "streaming.h"
  14 #include "thread-utils.h"
  15 #include "packfile.h"
  16
  17 static const char index_pack_usage[] =
  18 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  19
  20 struct object_entry {
  21         struct pack_idx_entry idx;
  22         unsigned long size;
  23         unsigned char hdr_size;
  24         signed char type;
  25         signed char real_type;
  26 };
  27
  28 struct object_stat {
  29         unsigned delta_depth;
  30         int base_object_no;
  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 /* Remember to update object flag allocation in object.h */
  53 #define FLAG_LINK (1u<<20)
  54 #define FLAG_CHECKED (1u<<21)
  55
  56 struct ofs_delta_entry {
  57         off_t offset;
  58         int obj_no;
  59 };
  60
  61 struct ref_delta_entry {
  62         unsigned char sha1[20];
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct object_stat *obj_stat;
  68 static struct ofs_delta_entry *ofs_deltas;
  69 static struct ref_delta_entry *ref_deltas;
  70 static struct thread_local nothread_data;
  71 static int nr_objects;
  72 static int nr_ofs_deltas;
  73 static int nr_ref_deltas;
  74 static int ref_deltas_alloc;
  75 static int nr_resolved_deltas;
  76 static int nr_threads;
  77
  78 static int from_stdin;
  79 static int strict;
  80 static int do_fsck_object;
  81 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  82 static int verbose;
  83 static int show_resolving_progress;
  84 static int show_stat;
  85 static int check_self_contained_and_connected;
  86
  87 static struct progress *progress;
  88
  89 /* We always read in 4kB chunks. */
  90 static unsigned char input_buffer[4096];
  91 static unsigned int input_offset, input_len;
  92 static off_t consumed_bytes;
  93 static off_t max_input_size;
  94 static unsigned deepest_delta;
  95 static git_hash_ctx input_ctx;
  96 static uint32_t input_crc32;
  97 static int input_fd, output_fd;
  98 static const char *curr_pack;
  99
 100 #ifndef NO_PTHREADS
 101
 102 static struct thread_local *thread_data;
 103 static int nr_dispatched;
 104 static int threads_active;
 105
 106 static pthread_mutex_t read_mutex;
 107 #define read_lock()             lock_mutex(&read_mutex)
 108 #define read_unlock()           unlock_mutex(&read_mutex)
 109
 110 static pthread_mutex_t counter_mutex;
 111 #define counter_lock()          lock_mutex(&counter_mutex)
 112 #define counter_unlock()        unlock_mutex(&counter_mutex)
 113
 114 static pthread_mutex_t work_mutex;
 115 #define work_lock()             lock_mutex(&work_mutex)
 116 #define work_unlock()           unlock_mutex(&work_mutex)
 117
 118 static pthread_mutex_t deepest_delta_mutex;
 119 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 120 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 121
 122 static pthread_mutex_t type_cas_mutex;
 123 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 124 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 125
 126 static pthread_key_t key;
 127
 128 static inline void lock_mutex(pthread_mutex_t *mutex)
 129 {
 130         if (threads_active)
 131                 pthread_mutex_lock(mutex);
 132 }
 133
 134 static inline void unlock_mutex(pthread_mutex_t *mutex)
 135 {
 136         if (threads_active)
 137                 pthread_mutex_unlock(mutex);
 138 }
 139
 140 /*
 141  * Mutex and conditional variable can't be statically-initialized on Windows.
 142  */
 143 static void init_thread(void)
 144 {
 145         int i;
 146         init_recursive_mutex(&read_mutex);
 147         pthread_mutex_init(&counter_mutex, NULL);
 148         pthread_mutex_init(&work_mutex, NULL);
 149         pthread_mutex_init(&type_cas_mutex, NULL);
 150         if (show_stat)
 151                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 152         pthread_key_create(&key, NULL);
 153         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 154         for (i = 0; i < nr_threads; i++) {
 155                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 156                 if (thread_data[i].pack_fd == -1)
 157                         die_errno(_("unable to open %s"), curr_pack);
 158         }
 159
 160         threads_active = 1;
 161 }
 162
 163 static void cleanup_thread(void)
 164 {
 165         int i;
 166         if (!threads_active)
 167                 return;
 168         threads_active = 0;
 169         pthread_mutex_destroy(&read_mutex);
 170         pthread_mutex_destroy(&counter_mutex);
 171         pthread_mutex_destroy(&work_mutex);
 172         pthread_mutex_destroy(&type_cas_mutex);
 173         if (show_stat)
 174                 pthread_mutex_destroy(&deepest_delta_mutex);
 175         for (i = 0; i < nr_threads; i++)
 176                 close(thread_data[i].pack_fd);
 177         pthread_key_delete(key);
 178         free(thread_data);
 179 }
 180
 181 #else
 182
 183 #define read_lock()
 184 #define read_unlock()
 185
 186 #define counter_lock()
 187 #define counter_unlock()
 188
 189 #define work_lock()
 190 #define work_unlock()
 191
 192 #define deepest_delta_lock()
 193 #define deepest_delta_unlock()
 194
 195 #define type_cas_lock()
 196 #define type_cas_unlock()
 197
 198 #endif
 199
 200
 201 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 202 {
 203         if (!obj)
 204                 return -1;
 205
 206         if (type != OBJ_ANY && obj->type != type)
 207                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 208
 209         obj->flags |= FLAG_LINK;
 210         return 0;
 211 }
 212
 213 /* The content of each linked object must have been checked
 214    or it must be already present in the object database */
 215 static unsigned check_object(struct object *obj)
 216 {
 217         if (!obj)
 218                 return 0;
 219
 220         if (!(obj->flags & FLAG_LINK))
 221                 return 0;
 222
 223         if (!(obj->flags & FLAG_CHECKED)) {
 224                 unsigned long size;
 225                 int type = sha1_object_info(obj->oid.hash, &size);
 226                 if (type <= 0)
 227                         die(_("did not receive expected object %s"),
 228                               oid_to_hex(&obj->oid));
 229                 if (type != obj->type)
 230                         die(_("object %s: expected type %s, found %s"),
 231                             oid_to_hex(&obj->oid),
 232                             type_name(obj->type), type_name(type));
 233                 obj->flags |= FLAG_CHECKED;
 234                 return 1;
 235         }
 236
 237         return 0;
 238 }
 239
 240 static unsigned check_objects(void)
 241 {
 242         unsigned i, max, foreign_nr = 0;
 243
 244         max = get_max_object_index();
 245         for (i = 0; i < max; i++)
 246                 foreign_nr += check_object(get_indexed_object(i));
 247         return foreign_nr;
 248 }
 249
 250
 251 /* Discard current buffer used content. */
 252 static void flush(void)
 253 {
 254         if (input_offset) {
 255                 if (output_fd >= 0)
 256                         write_or_die(output_fd, input_buffer, input_offset);
 257                 the_hash_algo->update_fn(&input_ctx, input_buffer, input_offset);
 258                 memmove(input_buffer, input_buffer + input_offset, input_len);
 259                 input_offset = 0;
 260         }
 261 }
 262
 263 /*
 264  * Make sure at least "min" bytes are available in the buffer, and
 265  * return the pointer to the buffer.
 266  */
 267 static void *fill(int min)
 268 {
 269         if (min <= input_len)
 270                 return input_buffer + input_offset;
 271         if (min > sizeof(input_buffer))
 272                 die(Q_("cannot fill %d byte",
 273                        "cannot fill %d bytes",
 274                        min),
 275                     min);
 276         flush();
 277         do {
 278                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 279                                 sizeof(input_buffer) - input_len);
 280                 if (ret <= 0) {
 281                         if (!ret)
 282                                 die(_("early EOF"));
 283                         die_errno(_("read error on input"));
 284                 }
 285                 input_len += ret;
 286                 if (from_stdin)
 287                         display_throughput(progress, consumed_bytes + input_len);
 288         } while (input_len < min);
 289         return input_buffer;
 290 }
 291
 292 static void use(int bytes)
 293 {
 294         if (bytes > input_len)
 295                 die(_("used more bytes than were available"));
 296         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 297         input_len -= bytes;
 298         input_offset += bytes;
 299
 300         /* make sure off_t is sufficiently large not to wrap */
 301         if (signed_add_overflows(consumed_bytes, bytes))
 302                 die(_("pack too large for current definition of off_t"));
 303         consumed_bytes += bytes;
 304         if (max_input_size && consumed_bytes > max_input_size)
 305                 die(_("pack exceeds maximum allowed size"));
 306 }
 307
 308 static const char *open_pack_file(const char *pack_name)
 309 {
 310         if (from_stdin) {
 311                 input_fd = 0;
 312                 if (!pack_name) {
 313                         struct strbuf tmp_file = STRBUF_INIT;
 314                         output_fd = odb_mkstemp(&tmp_file,
 315                                                 "pack/tmp_pack_XXXXXX");
 316                         pack_name = strbuf_detach(&tmp_file, NULL);
 317                 } else {
 318                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 319                         if (output_fd < 0)
 320                                 die_errno(_("unable to create '%s'"), pack_name);
 321                 }
 322                 nothread_data.pack_fd = output_fd;
 323         } else {
 324                 input_fd = open(pack_name, O_RDONLY);
 325                 if (input_fd < 0)
 326                         die_errno(_("cannot open packfile '%s'"), pack_name);
 327                 output_fd = -1;
 328                 nothread_data.pack_fd = input_fd;
 329         }
 330         the_hash_algo->init_fn(&input_ctx);
 331         return pack_name;
 332 }
 333
 334 static void parse_pack_header(void)
 335 {
 336         struct pack_header *hdr = fill(sizeof(struct pack_header));
 337
 338         /* Header consistency check */
 339         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 340                 die(_("pack signature mismatch"));
 341         if (!pack_version_ok(hdr->hdr_version))
 342                 die(_("pack version %"PRIu32" unsupported"),
 343                         ntohl(hdr->hdr_version));
 344
 345         nr_objects = ntohl(hdr->hdr_entries);
 346         use(sizeof(struct pack_header));
 347 }
 348
 349 static NORETURN void bad_object(off_t offset, const char *format,
 350                        ...) __attribute__((format (printf, 2, 3)));
 351
 352 static NORETURN void bad_object(off_t offset, const char *format, ...)
 353 {
 354         va_list params;
 355         char buf[1024];
 356
 357         va_start(params, format);
 358         vsnprintf(buf, sizeof(buf), format, params);
 359         va_end(params);
 360         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 361             (uintmax_t)offset, buf);
 362 }
 363
 364 static inline struct thread_local *get_thread_data(void)
 365 {
 366 #ifndef NO_PTHREADS
 367         if (threads_active)
 368                 return pthread_getspecific(key);
 369         assert(!threads_active &&
 370                "This should only be reached when all threads are gone");
 371 #endif
 372         return &nothread_data;
 373 }
 374
 375 #ifndef NO_PTHREADS
 376 static void set_thread_data(struct thread_local *data)
 377 {
 378         if (threads_active)
 379                 pthread_setspecific(key, data);
 380 }
 381 #endif
 382
 383 static struct base_data *alloc_base_data(void)
 384 {
 385         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 386         base->ref_last = -1;
 387         base->ofs_last = -1;
 388         return base;
 389 }
 390
 391 static void free_base_data(struct base_data *c)
 392 {
 393         if (c->data) {
 394                 FREE_AND_NULL(c->data);
 395                 get_thread_data()->base_cache_used -= c->size;
 396         }
 397 }
 398
 399 static void prune_base_data(struct base_data *retain)
 400 {
 401         struct base_data *b;
 402         struct thread_local *data = get_thread_data();
 403         for (b = data->base_cache;
 404              data->base_cache_used > delta_base_cache_limit && b;
 405              b = b->child) {
 406                 if (b->data && b != retain)
 407                         free_base_data(b);
 408         }
 409 }
 410
 411 static void link_base_data(struct base_data *base, struct base_data *c)
 412 {
 413         if (base)
 414                 base->child = c;
 415         else
 416                 get_thread_data()->base_cache = c;
 417
 418         c->base = base;
 419         c->child = NULL;
 420         if (c->data)
 421                 get_thread_data()->base_cache_used += c->size;
 422         prune_base_data(c);
 423 }
 424
 425 static void unlink_base_data(struct base_data *c)
 426 {
 427         struct base_data *base = c->base;
 428         if (base)
 429                 base->child = NULL;
 430         else
 431                 get_thread_data()->base_cache = NULL;
 432         free_base_data(c);
 433 }
 434
 435 static int is_delta_type(enum object_type type)
 436 {
 437         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 438 }
 439
 440 static void *unpack_entry_data(off_t offset, unsigned long size,
 441                                enum object_type type, struct object_id *oid)
 442 {
 443         static char fixed_buf[8192];
 444         int status;
 445         git_zstream stream;
 446         void *buf;
 447         git_hash_ctx c;
 448         char hdr[32];
 449         int hdrlen;
 450
 451         if (!is_delta_type(type)) {
 452                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", type_name(type), size) + 1;
 453                 the_hash_algo->init_fn(&c);
 454                 the_hash_algo->update_fn(&c, hdr, hdrlen);
 455         } else
 456                 oid = NULL;
 457         if (type == OBJ_BLOB && size > big_file_threshold)
 458                 buf = fixed_buf;
 459         else
 460                 buf = xmallocz(size);
 461
 462         memset(&stream, 0, sizeof(stream));
 463         git_inflate_init(&stream);
 464         stream.next_out = buf;
 465         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 466
 467         do {
 468                 unsigned char *last_out = stream.next_out;
 469                 stream.next_in = fill(1);
 470                 stream.avail_in = input_len;
 471                 status = git_inflate(&stream, 0);
 472                 use(input_len - stream.avail_in);
 473                 if (oid)
 474                         the_hash_algo->update_fn(&c, last_out, stream.next_out - last_out);
 475                 if (buf == fixed_buf) {
 476                         stream.next_out = buf;
 477                         stream.avail_out = sizeof(fixed_buf);
 478                 }
 479         } while (status == Z_OK);
 480         if (stream.total_out != size || status != Z_STREAM_END)
 481                 bad_object(offset, _("inflate returned %d"), status);
 482         git_inflate_end(&stream);
 483         if (oid)
 484                 the_hash_algo->final_fn(oid->hash, &c);
 485         return buf == fixed_buf ? NULL : buf;
 486 }
 487
 488 static void *unpack_raw_entry(struct object_entry *obj,
 489                               off_t *ofs_offset,
 490                               struct object_id *ref_oid,
 491                               struct object_id *oid)
 492 {
 493         unsigned char *p;
 494         unsigned long size, c;
 495         off_t base_offset;
 496         unsigned shift;
 497         void *data;
 498
 499         obj->idx.offset = consumed_bytes;
 500         input_crc32 = crc32(0, NULL, 0);
 501
 502         p = fill(1);
 503         c = *p;
 504         use(1);
 505         obj->type = (c >> 4) & 7;
 506         size = (c & 15);
 507         shift = 4;
 508         while (c & 0x80) {
 509                 p = fill(1);
 510                 c = *p;
 511                 use(1);
 512                 size += (c & 0x7f) << shift;
 513                 shift += 7;
 514         }
 515         obj->size = size;
 516
 517         switch (obj->type) {
 518         case OBJ_REF_DELTA:
 519                 hashcpy(ref_oid->hash, fill(the_hash_algo->rawsz));
 520                 use(the_hash_algo->rawsz);
 521                 break;
 522         case OBJ_OFS_DELTA:
 523                 p = fill(1);
 524                 c = *p;
 525                 use(1);
 526                 base_offset = c & 127;
 527                 while (c & 128) {
 528                         base_offset += 1;
 529                         if (!base_offset || MSB(base_offset, 7))
 530                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 531                         p = fill(1);
 532                         c = *p;
 533                         use(1);
 534                         base_offset = (base_offset << 7) + (c & 127);
 535                 }
 536                 *ofs_offset = obj->idx.offset - base_offset;
 537                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 538                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 539                 break;
 540         case OBJ_COMMIT:
 541         case OBJ_TREE:
 542         case OBJ_BLOB:
 543         case OBJ_TAG:
 544                 break;
 545         default:
 546                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 547         }
 548         obj->hdr_size = consumed_bytes - obj->idx.offset;
 549
 550         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, oid);
 551         obj->idx.crc32 = input_crc32;
 552         return data;
 553 }
 554
 555 static void *unpack_data(struct object_entry *obj,
 556                          int (*consume)(const unsigned char *, unsigned long, void *),
 557                          void *cb_data)
 558 {
 559         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 560         off_t len = obj[1].idx.offset - from;
 561         unsigned char *data, *inbuf;
 562         git_zstream stream;
 563         int status;
 564
 565         data = xmallocz(consume ? 64*1024 : obj->size);
 566         inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
 567
 568         memset(&stream, 0, sizeof(stream));
 569         git_inflate_init(&stream);
 570         stream.next_out = data;
 571         stream.avail_out = consume ? 64*1024 : obj->size;
 572
 573         do {
 574                 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
 575                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 576                 if (n < 0)
 577                         die_errno(_("cannot pread pack file"));
 578                 if (!n)
 579                         die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
 580                                "premature end of pack file, %"PRIuMAX" bytes missing",
 581                                (unsigned int)len),
 582                             (uintmax_t)len);
 583                 from += n;
 584                 len -= n;
 585                 stream.next_in = inbuf;
 586                 stream.avail_in = n;
 587                 if (!consume)
 588                         status = git_inflate(&stream, 0);
 589                 else {
 590                         do {
 591                                 status = git_inflate(&stream, 0);
 592                                 if (consume(data, stream.next_out - data, cb_data)) {
 593                                         free(inbuf);
 594                                         free(data);
 595                                         return NULL;
 596                                 }
 597                                 stream.next_out = data;
 598                                 stream.avail_out = 64*1024;
 599                         } while (status == Z_OK && stream.avail_in);
 600                 }
 601         } while (len && status == Z_OK && !stream.avail_in);
 602
 603         /* This has been inflated OK when first encountered, so... */
 604         if (status != Z_STREAM_END || stream.total_out != obj->size)
 605                 die(_("serious inflate inconsistency"));
 606
 607         git_inflate_end(&stream);
 608         free(inbuf);
 609         if (consume) {
 610                 FREE_AND_NULL(data);
 611         }
 612         return data;
 613 }
 614
 615 static void *get_data_from_pack(struct object_entry *obj)
 616 {
 617         return unpack_data(obj, NULL, NULL);
 618 }
 619
 620 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 621                                    enum object_type type1,
 622                                    enum object_type type2)
 623 {
 624         int cmp = type1 - type2;
 625         if (cmp)
 626                 return cmp;
 627         return offset1 < offset2 ? -1 :
 628                offset1 > offset2 ?  1 :
 629                0;
 630 }
 631
 632 static int find_ofs_delta(const off_t offset, enum object_type type)
 633 {
 634         int first = 0, last = nr_ofs_deltas;
 635
 636         while (first < last) {
 637                 int next = first + (last - first) / 2;
 638                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 639                 int cmp;
 640
 641                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 642                                               type, objects[delta->obj_no].type);
 643                 if (!cmp)
 644                         return next;
 645                 if (cmp < 0) {
 646                         last = next;
 647                         continue;
 648                 }
 649                 first = next+1;
 650         }
 651         return -first-1;
 652 }
 653
 654 static void find_ofs_delta_children(off_t offset,
 655                                     int *first_index, int *last_index,
 656                                     enum object_type type)
 657 {
 658         int first = find_ofs_delta(offset, type);
 659         int last = first;
 660         int end = nr_ofs_deltas - 1;
 661
 662         if (first < 0) {
 663                 *first_index = 0;
 664                 *last_index = -1;
 665                 return;
 666         }
 667         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 668                 --first;
 669         while (last < end && ofs_deltas[last + 1].offset == offset)
 670                 ++last;
 671         *first_index = first;
 672         *last_index = last;
 673 }
 674
 675 static int compare_ref_delta_bases(const unsigned char *sha1,
 676                                    const unsigned char *sha2,
 677                                    enum object_type type1,
 678                                    enum object_type type2)
 679 {
 680         int cmp = type1 - type2;
 681         if (cmp)
 682                 return cmp;
 683         return hashcmp(sha1, sha2);
 684 }
 685
 686 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
 687 {
 688         int first = 0, last = nr_ref_deltas;
 689
 690         while (first < last) {
 691                 int next = first + (last - first) / 2;
 692                 struct ref_delta_entry *delta = &ref_deltas[next];
 693                 int cmp;
 694
 695                 cmp = compare_ref_delta_bases(sha1, delta->sha1,
 696                                               type, objects[delta->obj_no].type);
 697                 if (!cmp)
 698                         return next;
 699                 if (cmp < 0) {
 700                         last = next;
 701                         continue;
 702                 }
 703                 first = next+1;
 704         }
 705         return -first-1;
 706 }
 707
 708 static void find_ref_delta_children(const unsigned char *sha1,
 709                                     int *first_index, int *last_index,
 710                                     enum object_type type)
 711 {
 712         int first = find_ref_delta(sha1, type);
 713         int last = first;
 714         int end = nr_ref_deltas - 1;
 715
 716         if (first < 0) {
 717                 *first_index = 0;
 718                 *last_index = -1;
 719                 return;
 720         }
 721         while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
 722                 --first;
 723         while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
 724                 ++last;
 725         *first_index = first;
 726         *last_index = last;
 727 }
 728
 729 struct compare_data {
 730         struct object_entry *entry;
 731         struct git_istream *st;
 732         unsigned char *buf;
 733         unsigned long buf_size;
 734 };
 735
 736 static int compare_objects(const unsigned char *buf, unsigned long size,
 737                            void *cb_data)
 738 {
 739         struct compare_data *data = cb_data;
 740
 741         if (data->buf_size < size) {
 742                 free(data->buf);
 743                 data->buf = xmalloc(size);
 744                 data->buf_size = size;
 745         }
 746
 747         while (size) {
 748                 ssize_t len = read_istream(data->st, data->buf, size);
 749                 if (len == 0)
 750                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 751                             oid_to_hex(&data->entry->idx.oid));
 752                 if (len < 0)
 753                         die(_("unable to read %s"),
 754                             oid_to_hex(&data->entry->idx.oid));
 755                 if (memcmp(buf, data->buf, len))
 756                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 757                             oid_to_hex(&data->entry->idx.oid));
 758                 size -= len;
 759                 buf += len;
 760         }
 761         return 0;
 762 }
 763
 764 static int check_collison(struct object_entry *entry)
 765 {
 766         struct compare_data data;
 767         enum object_type type;
 768         unsigned long size;
 769
 770         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 771                 return -1;
 772
 773         memset(&data, 0, sizeof(data));
 774         data.entry = entry;
 775         data.st = open_istream(entry->idx.oid.hash, &type, &size, NULL);
 776         if (!data.st)
 777                 return -1;
 778         if (size != entry->size || type != entry->type)
 779                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 780                     oid_to_hex(&entry->idx.oid));
 781         unpack_data(entry, compare_objects, &data);
 782         close_istream(data.st);
 783         free(data.buf);
 784         return 0;
 785 }
 786
 787 static void sha1_object(const void *data, struct object_entry *obj_entry,
 788                         unsigned long size, enum object_type type,
 789                         const struct object_id *oid)
 790 {
 791         void *new_data = NULL;
 792         int collision_test_needed = 0;
 793
 794         assert(data || obj_entry);
 795
 796         if (startup_info->have_repository) {
 797                 read_lock();
 798                 collision_test_needed =
 799                         has_sha1_file_with_flags(oid->hash, OBJECT_INFO_QUICK);
 800                 read_unlock();
 801         }
 802
 803         if (collision_test_needed && !data) {
 804                 read_lock();
 805                 if (!check_collison(obj_entry))
 806                         collision_test_needed = 0;
 807                 read_unlock();
 808         }
 809         if (collision_test_needed) {
 810                 void *has_data;
 811                 enum object_type has_type;
 812                 unsigned long has_size;
 813                 read_lock();
 814                 has_type = sha1_object_info(oid->hash, &has_size);
 815                 if (has_type < 0)
 816                         die(_("cannot read existing object info %s"), oid_to_hex(oid));
 817                 if (has_type != type || has_size != size)
 818                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 819                 has_data = read_sha1_file(oid->hash, &has_type, &has_size);
 820                 read_unlock();
 821                 if (!data)
 822                         data = new_data = get_data_from_pack(obj_entry);
 823                 if (!has_data)
 824                         die(_("cannot read existing object %s"), oid_to_hex(oid));
 825                 if (size != has_size || type != has_type ||
 826                     memcmp(data, has_data, size) != 0)
 827                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 828                 free(has_data);
 829         }
 830
 831         if (strict || do_fsck_object) {
 832                 read_lock();
 833                 if (type == OBJ_BLOB) {
 834                         struct blob *blob = lookup_blob(oid);
 835                         if (blob)
 836                                 blob->object.flags |= FLAG_CHECKED;
 837                         else
 838                                 die(_("invalid blob object %s"), oid_to_hex(oid));
 839                 } else {
 840                         struct object *obj;
 841                         int eaten;
 842                         void *buf = (void *) data;
 843
 844                         assert(data && "data can only be NULL for large _blobs_");
 845
 846                         /*
 847                          * we do not need to free the memory here, as the
 848                          * buf is deleted by the caller.
 849                          */
 850                         obj = parse_object_buffer(oid, type, size, buf,
 851                                                   &eaten);
 852                         if (!obj)
 853                                 die(_("invalid %s"), type_name(type));
 854                         if (do_fsck_object &&
 855                             fsck_object(obj, buf, size, &fsck_options))
 856                                 die(_("Error in object"));
 857                         if (strict && fsck_walk(obj, NULL, &fsck_options))
 858                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 859
 860                         if (obj->type == OBJ_TREE) {
 861                                 struct tree *item = (struct tree *) obj;
 862                                 item->buffer = NULL;
 863                                 obj->parsed = 0;
 864                         }
 865                         if (obj->type == OBJ_COMMIT) {
 866                                 struct commit *commit = (struct commit *) obj;
 867                                 if (detach_commit_buffer(commit, NULL) != data)
 868                                         die("BUG: parse_object_buffer transmogrified our buffer");
 869                         }
 870                         obj->flags |= FLAG_CHECKED;
 871                 }
 872                 read_unlock();
 873         }
 874
 875         free(new_data);
 876 }
 877
 878 /*
 879  * This function is part of find_unresolved_deltas(). There are two
 880  * walkers going in the opposite ways.
 881  *
 882  * The first one in find_unresolved_deltas() traverses down from
 883  * parent node to children, deflating nodes along the way. However,
 884  * memory for deflated nodes is limited by delta_base_cache_limit, so
 885  * at some point parent node's deflated content may be freed.
 886  *
 887  * The second walker is this function, which goes from current node up
 888  * to top parent if necessary to deflate the node. In normal
 889  * situation, its parent node would be already deflated, so it just
 890  * needs to apply delta.
 891  *
 892  * In the worst case scenario, parent node is no longer deflated because
 893  * we're running out of delta_base_cache_limit; we need to re-deflate
 894  * parents, possibly up to the top base.
 895  *
 896  * All deflated objects here are subject to be freed if we exceed
 897  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 898  * just need to make sure the last node is not freed.
 899  */
 900 static void *get_base_data(struct base_data *c)
 901 {
 902         if (!c->data) {
 903                 struct object_entry *obj = c->obj;
 904                 struct base_data **delta = NULL;
 905                 int delta_nr = 0, delta_alloc = 0;
 906
 907                 while (is_delta_type(c->obj->type) && !c->data) {
 908                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 909                         delta[delta_nr++] = c;
 910                         c = c->base;
 911                 }
 912                 if (!delta_nr) {
 913                         c->data = get_data_from_pack(obj);
 914                         c->size = obj->size;
 915                         get_thread_data()->base_cache_used += c->size;
 916                         prune_base_data(c);
 917                 }
 918                 for (; delta_nr > 0; delta_nr--) {
 919                         void *base, *raw;
 920                         c = delta[delta_nr - 1];
 921                         obj = c->obj;
 922                         base = get_base_data(c->base);
 923                         raw = get_data_from_pack(obj);
 924                         c->data = patch_delta(
 925                                 base, c->base->size,
 926                                 raw, obj->size,
 927                                 &c->size);
 928                         free(raw);
 929                         if (!c->data)
 930                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 931                         get_thread_data()->base_cache_used += c->size;
 932                         prune_base_data(c);
 933                 }
 934                 free(delta);
 935         }
 936         return c->data;
 937 }
 938
 939 static void resolve_delta(struct object_entry *delta_obj,
 940                           struct base_data *base, struct base_data *result)
 941 {
 942         void *base_data, *delta_data;
 943
 944         if (show_stat) {
 945                 int i = delta_obj - objects;
 946                 int j = base->obj - objects;
 947                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 948                 deepest_delta_lock();
 949                 if (deepest_delta < obj_stat[i].delta_depth)
 950                         deepest_delta = obj_stat[i].delta_depth;
 951                 deepest_delta_unlock();
 952                 obj_stat[i].base_object_no = j;
 953         }
 954         delta_data = get_data_from_pack(delta_obj);
 955         base_data = get_base_data(base);
 956         result->obj = delta_obj;
 957         result->data = patch_delta(base_data, base->size,
 958                                    delta_data, delta_obj->size, &result->size);
 959         free(delta_data);
 960         if (!result->data)
 961                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 962         hash_object_file(result->data, result->size,
 963                          type_name(delta_obj->real_type), &delta_obj->idx.oid);
 964         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 965                     &delta_obj->idx.oid);
 966         counter_lock();
 967         nr_resolved_deltas++;
 968         counter_unlock();
 969 }
 970
 971 /*
 972  * Standard boolean compare-and-swap: atomically check whether "*type" is
 973  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 974  * and return false.
 975  */
 976 static int compare_and_swap_type(signed char *type,
 977                                  enum object_type want,
 978                                  enum object_type set)
 979 {
 980         enum object_type old;
 981
 982         type_cas_lock();
 983         old = *type;
 984         if (old == want)
 985                 *type = set;
 986         type_cas_unlock();
 987
 988         return old == want;
 989 }
 990
 991 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 992                                                   struct base_data *prev_base)
 993 {
 994         if (base->ref_last == -1 && base->ofs_last == -1) {
 995                 find_ref_delta_children(base->obj->idx.oid.hash,
 996                                         &base->ref_first, &base->ref_last,
 997                                         OBJ_REF_DELTA);
 998
 999                 find_ofs_delta_children(base->obj->idx.offset,
1000                                         &base->ofs_first, &base->ofs_last,
1001                                         OBJ_OFS_DELTA);
1002
1003                 if (base->ref_last == -1 && base->ofs_last == -1) {
1004                         free(base->data);
1005                         return NULL;
1006                 }
1007
1008                 link_base_data(prev_base, base);
1009         }
1010
1011         if (base->ref_first <= base->ref_last) {
1012                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1013                 struct base_data *result = alloc_base_data();
1014
1015                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1016                                            base->obj->real_type))
1017                         die("BUG: child->real_type != OBJ_REF_DELTA");
1018
1019                 resolve_delta(child, base, result);
1020                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1021                         free_base_data(base);
1022
1023                 base->ref_first++;
1024                 return result;
1025         }
1026
1027         if (base->ofs_first <= base->ofs_last) {
1028                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1029                 struct base_data *result = alloc_base_data();
1030
1031                 assert(child->real_type == OBJ_OFS_DELTA);
1032                 child->real_type = base->obj->real_type;
1033                 resolve_delta(child, base, result);
1034                 if (base->ofs_first == base->ofs_last)
1035                         free_base_data(base);
1036
1037                 base->ofs_first++;
1038                 return result;
1039         }
1040
1041         unlink_base_data(base);
1042         return NULL;
1043 }
1044
1045 static void find_unresolved_deltas(struct base_data *base)
1046 {
1047         struct base_data *new_base, *prev_base = NULL;
1048         for (;;) {
1049                 new_base = find_unresolved_deltas_1(base, prev_base);
1050
1051                 if (new_base) {
1052                         prev_base = base;
1053                         base = new_base;
1054                 } else {
1055                         free(base);
1056                         base = prev_base;
1057                         if (!base)
1058                                 return;
1059                         prev_base = base->base;
1060                 }
1061         }
1062 }
1063
1064 static int compare_ofs_delta_entry(const void *a, const void *b)
1065 {
1066         const struct ofs_delta_entry *delta_a = a;
1067         const struct ofs_delta_entry *delta_b = b;
1068
1069         return delta_a->offset < delta_b->offset ? -1 :
1070                delta_a->offset > delta_b->offset ?  1 :
1071                0;
1072 }
1073
1074 static int compare_ref_delta_entry(const void *a, const void *b)
1075 {
1076         const struct ref_delta_entry *delta_a = a;
1077         const struct ref_delta_entry *delta_b = b;
1078
1079         return hashcmp(delta_a->sha1, delta_b->sha1);
1080 }
1081
1082 static void resolve_base(struct object_entry *obj)
1083 {
1084         struct base_data *base_obj = alloc_base_data();
1085         base_obj->obj = obj;
1086         base_obj->data = NULL;
1087         find_unresolved_deltas(base_obj);
1088 }
1089
1090 #ifndef NO_PTHREADS
1091 static void *threaded_second_pass(void *data)
1092 {
1093         set_thread_data(data);
1094         for (;;) {
1095                 int i;
1096                 counter_lock();
1097                 display_progress(progress, nr_resolved_deltas);
1098                 counter_unlock();
1099                 work_lock();
1100                 while (nr_dispatched < nr_objects &&
1101                        is_delta_type(objects[nr_dispatched].type))
1102                         nr_dispatched++;
1103                 if (nr_dispatched >= nr_objects) {
1104                         work_unlock();
1105                         break;
1106                 }
1107                 i = nr_dispatched++;
1108                 work_unlock();
1109
1110                 resolve_base(&objects[i]);
1111         }
1112         return NULL;
1113 }
1114 #endif
1115
1116 /*
1117  * First pass:
1118  * - find locations of all objects;
1119  * - calculate SHA1 of all non-delta objects;
1120  * - remember base (SHA1 or offset) for all deltas.
1121  */
1122 static void parse_pack_objects(unsigned char *hash)
1123 {
1124         int i, nr_delays = 0;
1125         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1126         struct object_id ref_delta_oid;
1127         struct stat st;
1128
1129         if (verbose)
1130                 progress = start_progress(
1131                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1132                                 nr_objects);
1133         for (i = 0; i < nr_objects; i++) {
1134                 struct object_entry *obj = &objects[i];
1135                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1136                                               &ref_delta_oid,
1137                                               &obj->idx.oid);
1138                 obj->real_type = obj->type;
1139                 if (obj->type == OBJ_OFS_DELTA) {
1140                         nr_ofs_deltas++;
1141                         ofs_delta->obj_no = i;
1142                         ofs_delta++;
1143                 } else if (obj->type == OBJ_REF_DELTA) {
1144                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1145                         hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_oid.hash);
1146                         ref_deltas[nr_ref_deltas].obj_no = i;
1147                         nr_ref_deltas++;
1148                 } else if (!data) {
1149                         /* large blobs, check later */
1150                         obj->real_type = OBJ_BAD;
1151                         nr_delays++;
1152                 } else
1153                         sha1_object(data, NULL, obj->size, obj->type,
1154                                     &obj->idx.oid);
1155                 free(data);
1156                 display_progress(progress, i+1);
1157         }
1158         objects[i].idx.offset = consumed_bytes;
1159         stop_progress(&progress);
1160
1161         /* Check pack integrity */
1162         flush();
1163         the_hash_algo->final_fn(hash, &input_ctx);
1164         if (hashcmp(fill(the_hash_algo->rawsz), hash))
1165                 die(_("pack is corrupted (SHA1 mismatch)"));
1166         use(the_hash_algo->rawsz);
1167
1168         /* If input_fd is a file, we should have reached its end now. */
1169         if (fstat(input_fd, &st))
1170                 die_errno(_("cannot fstat packfile"));
1171         if (S_ISREG(st.st_mode) &&
1172                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1173                 die(_("pack has junk at the end"));
1174
1175         for (i = 0; i < nr_objects; i++) {
1176                 struct object_entry *obj = &objects[i];
1177                 if (obj->real_type != OBJ_BAD)
1178                         continue;
1179                 obj->real_type = obj->type;
1180                 sha1_object(NULL, obj, obj->size, obj->type,
1181                             &obj->idx.oid);
1182                 nr_delays--;
1183         }
1184         if (nr_delays)
1185                 die(_("confusion beyond insanity in parse_pack_objects()"));
1186 }
1187
1188 /*
1189  * Second pass:
1190  * - for all non-delta objects, look if it is used as a base for
1191  *   deltas;
1192  * - if used as a base, uncompress the object and apply all deltas,
1193  *   recursively checking if the resulting object is used as a base
1194  *   for some more deltas.
1195  */
1196 static void resolve_deltas(void)
1197 {
1198         int i;
1199
1200         if (!nr_ofs_deltas && !nr_ref_deltas)
1201                 return;
1202
1203         /* Sort deltas by base SHA1/offset for fast searching */
1204         QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1205         QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1206
1207         if (verbose || show_resolving_progress)
1208                 progress = start_progress(_("Resolving deltas"),
1209                                           nr_ref_deltas + nr_ofs_deltas);
1210
1211 #ifndef NO_PTHREADS
1212         nr_dispatched = 0;
1213         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1214                 init_thread();
1215                 for (i = 0; i < nr_threads; i++) {
1216                         int ret = pthread_create(&thread_data[i].thread, NULL,
1217                                                  threaded_second_pass, thread_data + i);
1218                         if (ret)
1219                                 die(_("unable to create thread: %s"),
1220                                     strerror(ret));
1221                 }
1222                 for (i = 0; i < nr_threads; i++)
1223                         pthread_join(thread_data[i].thread, NULL);
1224                 cleanup_thread();
1225                 return;
1226         }
1227 #endif
1228
1229         for (i = 0; i < nr_objects; i++) {
1230                 struct object_entry *obj = &objects[i];
1231
1232                 if (is_delta_type(obj->type))
1233                         continue;
1234                 resolve_base(obj);
1235                 display_progress(progress, nr_resolved_deltas);
1236         }
1237 }
1238
1239 /*
1240  * Third pass:
1241  * - append objects to convert thin pack to full pack if required
1242  * - write the final pack hash
1243  */
1244 static void fix_unresolved_deltas(struct hashfile *f);
1245 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_hash)
1246 {
1247         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1248                 stop_progress(&progress);
1249                 /* Flush remaining pack final hash. */
1250                 flush();
1251                 return;
1252         }
1253
1254         if (fix_thin_pack) {
1255                 struct hashfile *f;
1256                 unsigned char read_hash[GIT_MAX_RAWSZ], tail_hash[GIT_MAX_RAWSZ];
1257                 struct strbuf msg = STRBUF_INIT;
1258                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1259                 int nr_objects_initial = nr_objects;
1260                 if (nr_unresolved <= 0)
1261                         die(_("confusion beyond insanity"));
1262                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1263                 memset(objects + nr_objects + 1, 0,
1264                        nr_unresolved * sizeof(*objects));
1265                 f = hashfd(output_fd, curr_pack);
1266                 fix_unresolved_deltas(f);
1267                 strbuf_addf(&msg, Q_("completed with %d local object",
1268                                      "completed with %d local objects",
1269                                      nr_objects - nr_objects_initial),
1270                             nr_objects - nr_objects_initial);
1271                 stop_progress_msg(&progress, msg.buf);
1272                 strbuf_release(&msg);
1273                 hashclose(f, tail_hash, 0);
1274                 hashcpy(read_hash, pack_hash);
1275                 fixup_pack_header_footer(output_fd, pack_hash,
1276                                          curr_pack, nr_objects,
1277                                          read_hash, consumed_bytes-the_hash_algo->rawsz);
1278                 if (hashcmp(read_hash, tail_hash) != 0)
1279                         die(_("Unexpected tail checksum for %s "
1280                               "(disk corruption?)"), curr_pack);
1281         }
1282         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1283                 die(Q_("pack has %d unresolved delta",
1284                        "pack has %d unresolved deltas",
1285                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1286                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1287 }
1288
1289 static int write_compressed(struct hashfile *f, void *in, unsigned int size)
1290 {
1291         git_zstream stream;
1292         int status;
1293         unsigned char outbuf[4096];
1294
1295         git_deflate_init(&stream, zlib_compression_level);
1296         stream.next_in = in;
1297         stream.avail_in = size;
1298
1299         do {
1300                 stream.next_out = outbuf;
1301                 stream.avail_out = sizeof(outbuf);
1302                 status = git_deflate(&stream, Z_FINISH);
1303                 hashwrite(f, outbuf, sizeof(outbuf) - stream.avail_out);
1304         } while (status == Z_OK);
1305
1306         if (status != Z_STREAM_END)
1307                 die(_("unable to deflate appended object (%d)"), status);
1308         size = stream.total_out;
1309         git_deflate_end(&stream);
1310         return size;
1311 }
1312
1313 static struct object_entry *append_obj_to_pack(struct hashfile *f,
1314                                const unsigned char *sha1, void *buf,
1315                                unsigned long size, enum object_type type)
1316 {
1317         struct object_entry *obj = &objects[nr_objects++];
1318         unsigned char header[10];
1319         unsigned long s = size;
1320         int n = 0;
1321         unsigned char c = (type << 4) | (s & 15);
1322         s >>= 4;
1323         while (s) {
1324                 header[n++] = c | 0x80;
1325                 c = s & 0x7f;
1326                 s >>= 7;
1327         }
1328         header[n++] = c;
1329         crc32_begin(f);
1330         hashwrite(f, header, n);
1331         obj[0].size = size;
1332         obj[0].hdr_size = n;
1333         obj[0].type = type;
1334         obj[0].real_type = type;
1335         obj[1].idx.offset = obj[0].idx.offset + n;
1336         obj[1].idx.offset += write_compressed(f, buf, size);
1337         obj[0].idx.crc32 = crc32_end(f);
1338         hashflush(f);
1339         hashcpy(obj->idx.oid.hash, sha1);
1340         return obj;
1341 }
1342
1343 static int delta_pos_compare(const void *_a, const void *_b)
1344 {
1345         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1346         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1347         return a->obj_no - b->obj_no;
1348 }
1349
1350 static void fix_unresolved_deltas(struct hashfile *f)
1351 {
1352         struct ref_delta_entry **sorted_by_pos;
1353         int i;
1354
1355         /*
1356          * Since many unresolved deltas may well be themselves base objects
1357          * for more unresolved deltas, we really want to include the
1358          * smallest number of base objects that would cover as much delta
1359          * as possible by picking the
1360          * trunc deltas first, allowing for other deltas to resolve without
1361          * additional base objects.  Since most base objects are to be found
1362          * before deltas depending on them, a good heuristic is to start
1363          * resolving deltas in the same order as their position in the pack.
1364          */
1365         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1366         for (i = 0; i < nr_ref_deltas; i++)
1367                 sorted_by_pos[i] = &ref_deltas[i];
1368         QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
1369
1370         for (i = 0; i < nr_ref_deltas; i++) {
1371                 struct ref_delta_entry *d = sorted_by_pos[i];
1372                 enum object_type type;
1373                 struct base_data *base_obj = alloc_base_data();
1374
1375                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1376                         continue;
1377                 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1378                 if (!base_obj->data)
1379                         continue;
1380
1381                 if (check_sha1_signature(d->sha1, base_obj->data,
1382                                 base_obj->size, type_name(type)))
1383                         die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1384                 base_obj->obj = append_obj_to_pack(f, d->sha1,
1385                                         base_obj->data, base_obj->size, type);
1386                 find_unresolved_deltas(base_obj);
1387                 display_progress(progress, nr_resolved_deltas);
1388         }
1389         free(sorted_by_pos);
1390 }
1391
1392 static const char *derive_filename(const char *pack_name, const char *suffix,
1393                                    struct strbuf *buf)
1394 {
1395         size_t len;
1396         if (!strip_suffix(pack_name, ".pack", &len))
1397                 die(_("packfile name '%s' does not end with '.pack'"),
1398                     pack_name);
1399         strbuf_add(buf, pack_name, len);
1400         strbuf_addch(buf, '.');
1401         strbuf_addstr(buf, suffix);
1402         return buf->buf;
1403 }
1404
1405 static void write_special_file(const char *suffix, const char *msg,
1406                                const char *pack_name, const unsigned char *hash,
1407                                const char **report)
1408 {
1409         struct strbuf name_buf = STRBUF_INIT;
1410         const char *filename;
1411         int fd;
1412         int msg_len = strlen(msg);
1413
1414         if (pack_name)
1415                 filename = derive_filename(pack_name, suffix, &name_buf);
1416         else
1417                 filename = odb_pack_name(&name_buf, hash, suffix);
1418
1419         fd = odb_pack_keep(filename);
1420         if (fd < 0) {
1421                 if (errno != EEXIST)
1422                         die_errno(_("cannot write %s file '%s'"),
1423                                   suffix, filename);
1424         } else {
1425                 if (msg_len > 0) {
1426                         write_or_die(fd, msg, msg_len);
1427                         write_or_die(fd, "\n", 1);
1428                 }
1429                 if (close(fd) != 0)
1430                         die_errno(_("cannot close written %s file '%s'"),
1431                                   suffix, filename);
1432                 if (report)
1433                         *report = suffix;
1434         }
1435         strbuf_release(&name_buf);
1436 }
1437
1438 static void final(const char *final_pack_name, const char *curr_pack_name,
1439                   const char *final_index_name, const char *curr_index_name,
1440                   const char *keep_msg, const char *promisor_msg,
1441                   unsigned char *hash)
1442 {
1443         const char *report = "pack";
1444         struct strbuf pack_name = STRBUF_INIT;
1445         struct strbuf index_name = STRBUF_INIT;
1446         int err;
1447
1448         if (!from_stdin) {
1449                 close(input_fd);
1450         } else {
1451                 fsync_or_die(output_fd, curr_pack_name);
1452                 err = close(output_fd);
1453                 if (err)
1454                         die_errno(_("error while closing pack file"));
1455         }
1456
1457         if (keep_msg)
1458                 write_special_file("keep", keep_msg, final_pack_name, hash,
1459                                    &report);
1460         if (promisor_msg)
1461                 write_special_file("promisor", promisor_msg, final_pack_name,
1462                                    hash, NULL);
1463
1464         if (final_pack_name != curr_pack_name) {
1465                 if (!final_pack_name)
1466                         final_pack_name = odb_pack_name(&pack_name, hash, "pack");
1467                 if (finalize_object_file(curr_pack_name, final_pack_name))
1468                         die(_("cannot store pack file"));
1469         } else if (from_stdin)
1470                 chmod(final_pack_name, 0444);
1471
1472         if (final_index_name != curr_index_name) {
1473                 if (!final_index_name)
1474                         final_index_name = odb_pack_name(&index_name, hash, "idx");
1475                 if (finalize_object_file(curr_index_name, final_index_name))
1476                         die(_("cannot store index file"));
1477         } else
1478                 chmod(final_index_name, 0444);
1479
1480         if (!from_stdin) {
1481                 printf("%s\n", sha1_to_hex(hash));
1482         } else {
1483                 struct strbuf buf = STRBUF_INIT;
1484
1485                 strbuf_addf(&buf, "%s\t%s\n", report, sha1_to_hex(hash));
1486                 write_or_die(1, buf.buf, buf.len);
1487                 strbuf_release(&buf);
1488
1489                 /*
1490                  * Let's just mimic git-unpack-objects here and write
1491                  * the last part of the input buffer to stdout.
1492                  */
1493                 while (input_len) {
1494                         err = xwrite(1, input_buffer + input_offset, input_len);
1495                         if (err <= 0)
1496                                 break;
1497                         input_len -= err;
1498                         input_offset += err;
1499                 }
1500         }
1501
1502         strbuf_release(&index_name);
1503         strbuf_release(&pack_name);
1504 }
1505
1506 static int git_index_pack_config(const char *k, const char *v, void *cb)
1507 {
1508         struct pack_idx_option *opts = cb;
1509
1510         if (!strcmp(k, "pack.indexversion")) {
1511                 opts->version = git_config_int(k, v);
1512                 if (opts->version > 2)
1513                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1514                 return 0;
1515         }
1516         if (!strcmp(k, "pack.threads")) {
1517                 nr_threads = git_config_int(k, v);
1518                 if (nr_threads < 0)
1519                         die(_("invalid number of threads specified (%d)"),
1520                             nr_threads);
1521 #ifdef NO_PTHREADS
1522                 if (nr_threads != 1)
1523                         warning(_("no threads support, ignoring %s"), k);
1524                 nr_threads = 1;
1525 #endif
1526                 return 0;
1527         }
1528         return git_default_config(k, v, cb);
1529 }
1530
1531 static int cmp_uint32(const void *a_, const void *b_)
1532 {
1533         uint32_t a = *((uint32_t *)a_);
1534         uint32_t b = *((uint32_t *)b_);
1535
1536         return (a < b) ? -1 : (a != b);
1537 }
1538
1539 static void read_v2_anomalous_offsets(struct packed_git *p,
1540                                       struct pack_idx_option *opts)
1541 {
1542         const uint32_t *idx1, *idx2;
1543         uint32_t i;
1544
1545         /* The address of the 4-byte offset table */
1546         idx1 = (((const uint32_t *)p->index_data)
1547                 + 2 /* 8-byte header */
1548                 + 256 /* fan out */
1549                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1550                 + p->num_objects /* CRC32 table */
1551                 );
1552
1553         /* The address of the 8-byte offset table */
1554         idx2 = idx1 + p->num_objects;
1555
1556         for (i = 0; i < p->num_objects; i++) {
1557                 uint32_t off = ntohl(idx1[i]);
1558                 if (!(off & 0x80000000))
1559                         continue;
1560                 off = off & 0x7fffffff;
1561                 check_pack_index_ptr(p, &idx2[off * 2]);
1562                 if (idx2[off * 2])
1563                         continue;
1564                 /*
1565                  * The real offset is ntohl(idx2[off * 2]) in high 4
1566                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1567                  * octets.  But idx2[off * 2] is Zero!!!
1568                  */
1569                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1570                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1571         }
1572
1573         QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32);
1574 }
1575
1576 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1577 {
1578         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1579
1580         if (!p)
1581                 die(_("Cannot open existing pack file '%s'"), pack_name);
1582         if (open_pack_index(p))
1583                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1584
1585         /* Read the attributes from the existing idx file */
1586         opts->version = p->index_version;
1587
1588         if (opts->version == 2)
1589                 read_v2_anomalous_offsets(p, opts);
1590
1591         /*
1592          * Get rid of the idx file as we do not need it anymore.
1593          * NEEDSWORK: extract this bit from free_pack_by_name() in
1594          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1595          * know we haven't installed this pack (hence we never have
1596          * read anything from it).
1597          */
1598         close_pack_index(p);
1599         free(p);
1600 }
1601
1602 static void show_pack_info(int stat_only)
1603 {
1604         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1605         unsigned long *chain_histogram = NULL;
1606
1607         if (deepest_delta)
1608                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1609
1610         for (i = 0; i < nr_objects; i++) {
1611                 struct object_entry *obj = &objects[i];
1612
1613                 if (is_delta_type(obj->type))
1614                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1615                 if (stat_only)
1616                         continue;
1617                 printf("%s %-6s %lu %lu %"PRIuMAX,
1618                        oid_to_hex(&obj->idx.oid),
1619                        type_name(obj->real_type), obj->size,
1620                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1621                        (uintmax_t)obj->idx.offset);
1622                 if (is_delta_type(obj->type)) {
1623                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1624                         printf(" %u %s", obj_stat[i].delta_depth,
1625                                oid_to_hex(&bobj->idx.oid));
1626                 }
1627                 putchar('\n');
1628         }
1629
1630         if (baseobjects)
1631                 printf_ln(Q_("non delta: %d object",
1632                              "non delta: %d objects",
1633                              baseobjects),
1634                           baseobjects);
1635         for (i = 0; i < deepest_delta; i++) {
1636                 if (!chain_histogram[i])
1637                         continue;
1638                 printf_ln(Q_("chain length = %d: %lu object",
1639                              "chain length = %d: %lu objects",
1640                              chain_histogram[i]),
1641                           i + 1,
1642                           chain_histogram[i]);
1643         }
1644 }
1645
1646 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1647 {
1648         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1649         const char *curr_index;
1650         const char *index_name = NULL, *pack_name = NULL;
1651         const char *keep_msg = NULL;
1652         const char *promisor_msg = NULL;
1653         struct strbuf index_name_buf = STRBUF_INIT;
1654         struct pack_idx_entry **idx_objects;
1655         struct pack_idx_option opts;
1656         unsigned char pack_hash[GIT_MAX_RAWSZ];
1657         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1658         int report_end_of_input = 0;
1659
1660         /*
1661          * index-pack never needs to fetch missing objects, since it only
1662          * accesses the repo to do hash collision checks
1663          */
1664         fetch_if_missing = 0;
1665
1666         if (argc == 2 && !strcmp(argv[1], "-h"))
1667                 usage(index_pack_usage);
1668
1669         check_replace_refs = 0;
1670         fsck_options.walk = mark_link;
1671
1672         reset_pack_idx_option(&opts);
1673         git_config(git_index_pack_config, &opts);
1674         if (prefix && chdir(prefix))
1675                 die(_("Cannot come back to cwd"));
1676
1677         for (i = 1; i < argc; i++) {
1678                 const char *arg = argv[i];
1679
1680                 if (*arg == '-') {
1681                         if (!strcmp(arg, "--stdin")) {
1682                                 from_stdin = 1;
1683                         } else if (!strcmp(arg, "--fix-thin")) {
1684                                 fix_thin_pack = 1;
1685                         } else if (skip_to_optional_arg(arg, "--strict", &arg)) {
1686                                 strict = 1;
1687                                 do_fsck_object = 1;
1688                                 fsck_set_msg_types(&fsck_options, arg);
1689                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1690                                 strict = 1;
1691                                 check_self_contained_and_connected = 1;
1692                         } else if (!strcmp(arg, "--fsck-objects")) {
1693                                 do_fsck_object = 1;
1694                         } else if (!strcmp(arg, "--verify")) {
1695                                 verify = 1;
1696                         } else if (!strcmp(arg, "--verify-stat")) {
1697                                 verify = 1;
1698                                 show_stat = 1;
1699                         } else if (!strcmp(arg, "--verify-stat-only")) {
1700                                 verify = 1;
1701                                 show_stat = 1;
1702                                 stat_only = 1;
1703                         } else if (skip_to_optional_arg(arg, "--keep", &keep_msg)) {
1704                                 ; /* nothing to do */
1705                         } else if (skip_to_optional_arg(arg, "--promisor", &promisor_msg)) {
1706                                 ; /* already parsed */
1707                         } else if (starts_with(arg, "--threads=")) {
1708                                 char *end;
1709                                 nr_threads = strtoul(arg+10, &end, 0);
1710                                 if (!arg[10] || *end || nr_threads < 0)
1711                                         usage(index_pack_usage);
1712 #ifdef NO_PTHREADS
1713                                 if (nr_threads != 1)
1714                                         warning(_("no threads support, "
1715                                                   "ignoring %s"), arg);
1716                                 nr_threads = 1;
1717 #endif
1718                         } else if (starts_with(arg, "--pack_header=")) {
1719                                 struct pack_header *hdr;
1720                                 char *c;
1721
1722                                 hdr = (struct pack_header *)input_buffer;
1723                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1724                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1725                                 if (*c != ',')
1726                                         die(_("bad %s"), arg);
1727                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1728                                 if (*c)
1729                                         die(_("bad %s"), arg);
1730                                 input_len = sizeof(*hdr);
1731                         } else if (!strcmp(arg, "-v")) {
1732                                 verbose = 1;
1733                         } else if (!strcmp(arg, "--show-resolving-progress")) {
1734                                 show_resolving_progress = 1;
1735                         } else if (!strcmp(arg, "--report-end-of-input")) {
1736                                 report_end_of_input = 1;
1737                         } else if (!strcmp(arg, "-o")) {
1738                                 if (index_name || (i+1) >= argc)
1739                                         usage(index_pack_usage);
1740                                 index_name = argv[++i];
1741                         } else if (starts_with(arg, "--index-version=")) {
1742                                 char *c;
1743                                 opts.version = strtoul(arg + 16, &c, 10);
1744                                 if (opts.version > 2)
1745                                         die(_("bad %s"), arg);
1746                                 if (*c == ',')
1747                                         opts.off32_limit = strtoul(c+1, &c, 0);
1748                                 if (*c || opts.off32_limit & 0x80000000)
1749                                         die(_("bad %s"), arg);
1750                         } else if (skip_prefix(arg, "--max-input-size=", &arg)) {
1751                                 max_input_size = strtoumax(arg, NULL, 10);
1752                         } else
1753                                 usage(index_pack_usage);
1754                         continue;
1755                 }
1756
1757                 if (pack_name)
1758                         usage(index_pack_usage);
1759                 pack_name = arg;
1760         }
1761
1762         if (!pack_name && !from_stdin)
1763                 usage(index_pack_usage);
1764         if (fix_thin_pack && !from_stdin)
1765                 die(_("--fix-thin cannot be used without --stdin"));
1766         if (from_stdin && !startup_info->have_repository)
1767                 die(_("--stdin requires a git repository"));
1768         if (!index_name && pack_name)
1769                 index_name = derive_filename(pack_name, "idx", &index_name_buf);
1770
1771         if (verify) {
1772                 if (!index_name)
1773                         die(_("--verify with no packfile name given"));
1774                 read_idx_option(&opts, index_name);
1775                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1776         }
1777         if (strict)
1778                 opts.flags |= WRITE_IDX_STRICT;
1779
1780 #ifndef NO_PTHREADS
1781         if (!nr_threads) {
1782                 nr_threads = online_cpus();
1783                 /* An experiment showed that more threads does not mean faster */
1784                 if (nr_threads > 3)
1785                         nr_threads = 3;
1786         }
1787 #endif
1788
1789         curr_pack = open_pack_file(pack_name);
1790         parse_pack_header();
1791         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1792         if (show_stat)
1793                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1794         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1795         parse_pack_objects(pack_hash);
1796         if (report_end_of_input)
1797                 write_in_full(2, "\0", 1);
1798         resolve_deltas();
1799         conclude_pack(fix_thin_pack, curr_pack, pack_hash);
1800         free(ofs_deltas);
1801         free(ref_deltas);
1802         if (strict)
1803                 foreign_nr = check_objects();
1804
1805         if (show_stat)
1806                 show_pack_info(stat_only);
1807
1808         ALLOC_ARRAY(idx_objects, nr_objects);
1809         for (i = 0; i < nr_objects; i++)
1810                 idx_objects[i] = &objects[i].idx;
1811         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_hash);
1812         free(idx_objects);
1813
1814         if (!verify)
1815                 final(pack_name, curr_pack,
1816                       index_name, curr_index,
1817                       keep_msg, promisor_msg,
1818                       pack_hash);
1819         else
1820                 close(input_fd);
1821         free(objects);
1822         strbuf_release(&index_name_buf);
1823         if (pack_name == NULL)
1824                 free((void *) curr_pack);
1825         if (index_name == NULL)
1826                 free((void *) curr_index);
1827
1828         /*
1829          * Let the caller know this pack is not self contained
1830          */
1831         if (check_self_contained_and_connected && foreign_nr)
1832                 return 1;
1833
1834         return 0;
1835 }