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