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

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned char hdr_size;
  22         signed char type;
  23         signed char real_type;
  24 };
  25
  26 struct object_stat {
  27         unsigned delta_depth;
  28         int base_object_no;
  29 };
  30
  31 struct base_data {
  32         struct base_data *base;
  33         struct base_data *child;
  34         struct object_entry *obj;
  35         void *data;
  36         unsigned long size;
  37         int ref_first, ref_last;
  38         int ofs_first, ofs_last;
  39 };
  40
  41 struct thread_local {
  42 #ifndef NO_PTHREADS
  43         pthread_t thread;
  44 #endif
  45         struct base_data *base_cache;
  46         size_t base_cache_used;
  47         int pack_fd;
  48 };
  49
  50 #define FLAG_LINK (1u<<20)
  51 #define FLAG_CHECKED (1u<<21)
  52
  53 struct ofs_delta_entry {
  54         off_t offset;
  55         int obj_no;
  56 };
  57
  58 struct ref_delta_entry {
  59         unsigned char sha1[20];
  60         int obj_no;
  61 };
  62
  63 static struct object_entry *objects;
  64 static struct object_stat *obj_stat;
  65 static struct ofs_delta_entry *ofs_deltas;
  66 static struct ref_delta_entry *ref_deltas;
  67 static struct thread_local nothread_data;
  68 static int nr_objects;
  69 static int nr_ofs_deltas;
  70 static int nr_ref_deltas;
  71 static int ref_deltas_alloc;
  72 static int nr_resolved_deltas;
  73 static int nr_threads;
  74
  75 static int from_stdin;
  76 static int strict;
  77 static int do_fsck_object;
  78 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  79 static int verbose;
  80 static int show_stat;
  81 static int check_self_contained_and_connected;
  82
  83 static struct progress *progress;
  84
  85 /* We always read in 4kB chunks. */
  86 static unsigned char input_buffer[4096];
  87 static unsigned int input_offset, input_len;
  88 static off_t consumed_bytes;
  89 static unsigned deepest_delta;
  90 static git_SHA_CTX input_ctx;
  91 static uint32_t input_crc32;
  92 static int input_fd, output_fd;
  93 static const char *curr_pack;
  94
  95 #ifndef NO_PTHREADS
  96
  97 static struct thread_local *thread_data;
  98 static int nr_dispatched;
  99 static int threads_active;
 100
 101 static pthread_mutex_t read_mutex;
 102 #define read_lock()             lock_mutex(&read_mutex)
 103 #define read_unlock()           unlock_mutex(&read_mutex)
 104
 105 static pthread_mutex_t counter_mutex;
 106 #define counter_lock()          lock_mutex(&counter_mutex)
 107 #define counter_unlock()        unlock_mutex(&counter_mutex)
 108
 109 static pthread_mutex_t work_mutex;
 110 #define work_lock()             lock_mutex(&work_mutex)
 111 #define work_unlock()           unlock_mutex(&work_mutex)
 112
 113 static pthread_mutex_t deepest_delta_mutex;
 114 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 115 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 116
 117 static pthread_mutex_t type_cas_mutex;
 118 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 119 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 120
 121 static pthread_key_t key;
 122
 123 static inline void lock_mutex(pthread_mutex_t *mutex)
 124 {
 125         if (threads_active)
 126                 pthread_mutex_lock(mutex);
 127 }
 128
 129 static inline void unlock_mutex(pthread_mutex_t *mutex)
 130 {
 131         if (threads_active)
 132                 pthread_mutex_unlock(mutex);
 133 }
 134
 135 /*
 136  * Mutex and conditional variable can't be statically-initialized on Windows.
 137  */
 138 static void init_thread(void)
 139 {
 140         int i;
 141         init_recursive_mutex(&read_mutex);
 142         pthread_mutex_init(&counter_mutex, NULL);
 143         pthread_mutex_init(&work_mutex, NULL);
 144         pthread_mutex_init(&type_cas_mutex, NULL);
 145         if (show_stat)
 146                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 147         pthread_key_create(&key, NULL);
 148         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 149         for (i = 0; i < nr_threads; i++) {
 150                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 151                 if (thread_data[i].pack_fd == -1)
 152                         die_errno(_("unable to open %s"), curr_pack);
 153         }
 154
 155         threads_active = 1;
 156 }
 157
 158 static void cleanup_thread(void)
 159 {
 160         int i;
 161         if (!threads_active)
 162                 return;
 163         threads_active = 0;
 164         pthread_mutex_destroy(&read_mutex);
 165         pthread_mutex_destroy(&counter_mutex);
 166         pthread_mutex_destroy(&work_mutex);
 167         pthread_mutex_destroy(&type_cas_mutex);
 168         if (show_stat)
 169                 pthread_mutex_destroy(&deepest_delta_mutex);
 170         for (i = 0; i < nr_threads; i++)
 171                 close(thread_data[i].pack_fd);
 172         pthread_key_delete(key);
 173         free(thread_data);
 174 }
 175
 176 #else
 177
 178 #define read_lock()
 179 #define read_unlock()
 180
 181 #define counter_lock()
 182 #define counter_unlock()
 183
 184 #define work_lock()
 185 #define work_unlock()
 186
 187 #define deepest_delta_lock()
 188 #define deepest_delta_unlock()
 189
 190 #define type_cas_lock()
 191 #define type_cas_unlock()
 192
 193 #endif
 194
 195
 196 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 197 {
 198         if (!obj)
 199                 return -1;
 200
 201         if (type != OBJ_ANY && obj->type != type)
 202                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 203
 204         obj->flags |= FLAG_LINK;
 205         return 0;
 206 }
 207
 208 /* The content of each linked object must have been checked
 209    or it must be already present in the object database */
 210 static unsigned check_object(struct object *obj)
 211 {
 212         if (!obj)
 213                 return 0;
 214
 215         if (!(obj->flags & FLAG_LINK))
 216                 return 0;
 217
 218         if (!(obj->flags & FLAG_CHECKED)) {
 219                 unsigned long size;
 220                 int type = sha1_object_info(obj->oid.hash, &size);
 221                 if (type <= 0)
 222                         die(_("did not receive expected object %s"),
 223                               oid_to_hex(&obj->oid));
 224                 if (type != obj->type)
 225                         die(_("object %s: expected type %s, found %s"),
 226                             oid_to_hex(&obj->oid),
 227                             typename(obj->type), typename(type));
 228                 obj->flags |= FLAG_CHECKED;
 229                 return 1;
 230         }
 231
 232         return 0;
 233 }
 234
 235 static unsigned check_objects(void)
 236 {
 237         unsigned i, max, foreign_nr = 0;
 238
 239         max = get_max_object_index();
 240         for (i = 0; i < max; i++)
 241                 foreign_nr += check_object(get_indexed_object(i));
 242         return foreign_nr;
 243 }
 244
 245
 246 /* Discard current buffer used content. */
 247 static void flush(void)
 248 {
 249         if (input_offset) {
 250                 if (output_fd >= 0)
 251                         write_or_die(output_fd, input_buffer, input_offset);
 252                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 253                 memmove(input_buffer, input_buffer + input_offset, input_len);
 254                 input_offset = 0;
 255         }
 256 }
 257
 258 /*
 259  * Make sure at least "min" bytes are available in the buffer, and
 260  * return the pointer to the buffer.
 261  */
 262 static void *fill(int min)
 263 {
 264         if (min <= input_len)
 265                 return input_buffer + input_offset;
 266         if (min > sizeof(input_buffer))
 267                 die(Q_("cannot fill %d byte",
 268                        "cannot fill %d bytes",
 269                        min),
 270                     min);
 271         flush();
 272         do {
 273                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 274                                 sizeof(input_buffer) - input_len);
 275                 if (ret <= 0) {
 276                         if (!ret)
 277                                 die(_("early EOF"));
 278                         die_errno(_("read error on input"));
 279                 }
 280                 input_len += ret;
 281                 if (from_stdin)
 282                         display_throughput(progress, consumed_bytes + input_len);
 283         } while (input_len < min);
 284         return input_buffer;
 285 }
 286
 287 static void use(int bytes)
 288 {
 289         if (bytes > input_len)
 290                 die(_("used more bytes than were available"));
 291         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 292         input_len -= bytes;
 293         input_offset += bytes;
 294
 295         /* make sure off_t is sufficiently large not to wrap */
 296         if (signed_add_overflows(consumed_bytes, bytes))
 297                 die(_("pack too large for current definition of off_t"));
 298         consumed_bytes += bytes;
 299 }
 300
 301 static const char *open_pack_file(const char *pack_name)
 302 {
 303         if (from_stdin) {
 304                 input_fd = 0;
 305                 if (!pack_name) {
 306                         static char tmp_file[PATH_MAX];
 307                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 308                                                 "pack/tmp_pack_XXXXXX");
 309                         pack_name = xstrdup(tmp_file);
 310                 } else
 311                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 312                 if (output_fd < 0)
 313                         die_errno(_("unable to create '%s'"), pack_name);
 314                 nothread_data.pack_fd = output_fd;
 315         } else {
 316                 input_fd = open(pack_name, O_RDONLY);
 317                 if (input_fd < 0)
 318                         die_errno(_("cannot open packfile '%s'"), pack_name);
 319                 output_fd = -1;
 320                 nothread_data.pack_fd = input_fd;
 321         }
 322         git_SHA1_Init(&input_ctx);
 323         return pack_name;
 324 }
 325
 326 static void parse_pack_header(void)
 327 {
 328         struct pack_header *hdr = fill(sizeof(struct pack_header));
 329
 330         /* Header consistency check */
 331         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 332                 die(_("pack signature mismatch"));
 333         if (!pack_version_ok(hdr->hdr_version))
 334                 die(_("pack version %"PRIu32" unsupported"),
 335                         ntohl(hdr->hdr_version));
 336
 337         nr_objects = ntohl(hdr->hdr_entries);
 338         use(sizeof(struct pack_header));
 339 }
 340
 341 static NORETURN void bad_object(unsigned long offset, const char *format,
 342                        ...) __attribute__((format (printf, 2, 3)));
 343
 344 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 345 {
 346         va_list params;
 347         char buf[1024];
 348
 349         va_start(params, format);
 350         vsnprintf(buf, sizeof(buf), format, params);
 351         va_end(params);
 352         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 353 }
 354
 355 static inline struct thread_local *get_thread_data(void)
 356 {
 357 #ifndef NO_PTHREADS
 358         if (threads_active)
 359                 return pthread_getspecific(key);
 360         assert(!threads_active &&
 361                "This should only be reached when all threads are gone");
 362 #endif
 363         return &nothread_data;
 364 }
 365
 366 #ifndef NO_PTHREADS
 367 static void set_thread_data(struct thread_local *data)
 368 {
 369         if (threads_active)
 370                 pthread_setspecific(key, data);
 371 }
 372 #endif
 373
 374 static struct base_data *alloc_base_data(void)
 375 {
 376         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 377         base->ref_last = -1;
 378         base->ofs_last = -1;
 379         return base;
 380 }
 381
 382 static void free_base_data(struct base_data *c)
 383 {
 384         if (c->data) {
 385                 free(c->data);
 386                 c->data = NULL;
 387                 get_thread_data()->base_cache_used -= c->size;
 388         }
 389 }
 390
 391 static void prune_base_data(struct base_data *retain)
 392 {
 393         struct base_data *b;
 394         struct thread_local *data = get_thread_data();
 395         for (b = data->base_cache;
 396              data->base_cache_used > delta_base_cache_limit && b;
 397              b = b->child) {
 398                 if (b->data && b != retain)
 399                         free_base_data(b);
 400         }
 401 }
 402
 403 static void link_base_data(struct base_data *base, struct base_data *c)
 404 {
 405         if (base)
 406                 base->child = c;
 407         else
 408                 get_thread_data()->base_cache = c;
 409
 410         c->base = base;
 411         c->child = NULL;
 412         if (c->data)
 413                 get_thread_data()->base_cache_used += c->size;
 414         prune_base_data(c);
 415 }
 416
 417 static void unlink_base_data(struct base_data *c)
 418 {
 419         struct base_data *base = c->base;
 420         if (base)
 421                 base->child = NULL;
 422         else
 423                 get_thread_data()->base_cache = NULL;
 424         free_base_data(c);
 425 }
 426
 427 static int is_delta_type(enum object_type type)
 428 {
 429         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 430 }
 431
 432 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 433                                enum object_type type, unsigned char *sha1)
 434 {
 435         static char fixed_buf[8192];
 436         int status;
 437         git_zstream stream;
 438         void *buf;
 439         git_SHA_CTX c;
 440         char hdr[32];
 441         int hdrlen;
 442
 443         if (!is_delta_type(type)) {
 444                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(type), size) + 1;
 445                 git_SHA1_Init(&c);
 446                 git_SHA1_Update(&c, hdr, hdrlen);
 447         } else
 448                 sha1 = NULL;
 449         if (type == OBJ_BLOB && size > big_file_threshold)
 450                 buf = fixed_buf;
 451         else
 452                 buf = xmallocz(size);
 453
 454         memset(&stream, 0, sizeof(stream));
 455         git_inflate_init(&stream);
 456         stream.next_out = buf;
 457         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 458
 459         do {
 460                 unsigned char *last_out = stream.next_out;
 461                 stream.next_in = fill(1);
 462                 stream.avail_in = input_len;
 463                 status = git_inflate(&stream, 0);
 464                 use(input_len - stream.avail_in);
 465                 if (sha1)
 466                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 467                 if (buf == fixed_buf) {
 468                         stream.next_out = buf;
 469                         stream.avail_out = sizeof(fixed_buf);
 470                 }
 471         } while (status == Z_OK);
 472         if (stream.total_out != size || status != Z_STREAM_END)
 473                 bad_object(offset, _("inflate returned %d"), status);
 474         git_inflate_end(&stream);
 475         if (sha1)
 476                 git_SHA1_Final(sha1, &c);
 477         return buf == fixed_buf ? NULL : buf;
 478 }
 479
 480 static void *unpack_raw_entry(struct object_entry *obj,
 481                               off_t *ofs_offset,
 482                               unsigned char *ref_sha1,
 483                               unsigned char *sha1)
 484 {
 485         unsigned char *p;
 486         unsigned long size, c;
 487         off_t base_offset;
 488         unsigned shift;
 489         void *data;
 490
 491         obj->idx.offset = consumed_bytes;
 492         input_crc32 = crc32(0, NULL, 0);
 493
 494         p = fill(1);
 495         c = *p;
 496         use(1);
 497         obj->type = (c >> 4) & 7;
 498         size = (c & 15);
 499         shift = 4;
 500         while (c & 0x80) {
 501                 p = fill(1);
 502                 c = *p;
 503                 use(1);
 504                 size += (c & 0x7f) << shift;
 505                 shift += 7;
 506         }
 507         obj->size = size;
 508
 509         switch (obj->type) {
 510         case OBJ_REF_DELTA:
 511                 hashcpy(ref_sha1, fill(20));
 512                 use(20);
 513                 break;
 514         case OBJ_OFS_DELTA:
 515                 p = fill(1);
 516                 c = *p;
 517                 use(1);
 518                 base_offset = c & 127;
 519                 while (c & 128) {
 520                         base_offset += 1;
 521                         if (!base_offset || MSB(base_offset, 7))
 522                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 523                         p = fill(1);
 524                         c = *p;
 525                         use(1);
 526                         base_offset = (base_offset << 7) + (c & 127);
 527                 }
 528                 *ofs_offset = obj->idx.offset - base_offset;
 529                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 530                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 531                 break;
 532         case OBJ_COMMIT:
 533         case OBJ_TREE:
 534         case OBJ_BLOB:
 535         case OBJ_TAG:
 536                 break;
 537         default:
 538                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 539         }
 540         obj->hdr_size = consumed_bytes - obj->idx.offset;
 541
 542         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 543         obj->idx.crc32 = input_crc32;
 544         return data;
 545 }
 546
 547 static void *unpack_data(struct object_entry *obj,
 548                          int (*consume)(const unsigned char *, unsigned long, void *),
 549                          void *cb_data)
 550 {
 551         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 552         unsigned long len = obj[1].idx.offset - from;
 553         unsigned char *data, *inbuf;
 554         git_zstream stream;
 555         int status;
 556
 557         data = xmallocz(consume ? 64*1024 : obj->size);
 558         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 559
 560         memset(&stream, 0, sizeof(stream));
 561         git_inflate_init(&stream);
 562         stream.next_out = data;
 563         stream.avail_out = consume ? 64*1024 : obj->size;
 564
 565         do {
 566                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 567                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 568                 if (n < 0)
 569                         die_errno(_("cannot pread pack file"));
 570                 if (!n)
 571                         die(Q_("premature end of pack file, %lu byte missing",
 572                                "premature end of pack file, %lu bytes missing",
 573                                len),
 574                             len);
 575                 from += n;
 576                 len -= n;
 577                 stream.next_in = inbuf;
 578                 stream.avail_in = n;
 579                 if (!consume)
 580                         status = git_inflate(&stream, 0);
 581                 else {
 582                         do {
 583                                 status = git_inflate(&stream, 0);
 584                                 if (consume(data, stream.next_out - data, cb_data)) {
 585                                         free(inbuf);
 586                                         free(data);
 587                                         return NULL;
 588                                 }
 589                                 stream.next_out = data;
 590                                 stream.avail_out = 64*1024;
 591                         } while (status == Z_OK && stream.avail_in);
 592                 }
 593         } while (len && status == Z_OK && !stream.avail_in);
 594
 595         /* This has been inflated OK when first encountered, so... */
 596         if (status != Z_STREAM_END || stream.total_out != obj->size)
 597                 die(_("serious inflate inconsistency"));
 598
 599         git_inflate_end(&stream);
 600         free(inbuf);
 601         if (consume) {
 602                 free(data);
 603                 data = NULL;
 604         }
 605         return data;
 606 }
 607
 608 static void *get_data_from_pack(struct object_entry *obj)
 609 {
 610         return unpack_data(obj, NULL, NULL);
 611 }
 612
 613 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 614                                    enum object_type type1,
 615                                    enum object_type type2)
 616 {
 617         int cmp = type1 - type2;
 618         if (cmp)
 619                 return cmp;
 620         return offset1 < offset2 ? -1 :
 621                offset1 > offset2 ?  1 :
 622                0;
 623 }
 624
 625 static int find_ofs_delta(const off_t offset, enum object_type type)
 626 {
 627         int first = 0, last = nr_ofs_deltas;
 628
 629         while (first < last) {
 630                 int next = (first + last) / 2;
 631                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 632                 int cmp;
 633
 634                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 635                                               type, objects[delta->obj_no].type);
 636                 if (!cmp)
 637                         return next;
 638                 if (cmp < 0) {
 639                         last = next;
 640                         continue;
 641                 }
 642                 first = next+1;
 643         }
 644         return -first-1;
 645 }
 646
 647 static void find_ofs_delta_children(off_t offset,
 648                                     int *first_index, int *last_index,
 649                                     enum object_type type)
 650 {
 651         int first = find_ofs_delta(offset, type);
 652         int last = first;
 653         int end = nr_ofs_deltas - 1;
 654
 655         if (first < 0) {
 656                 *first_index = 0;
 657                 *last_index = -1;
 658                 return;
 659         }
 660         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 661                 --first;
 662         while (last < end && ofs_deltas[last + 1].offset == offset)
 663                 ++last;
 664         *first_index = first;
 665         *last_index = last;
 666 }
 667
 668 static int compare_ref_delta_bases(const unsigned char *sha1,
 669                                    const unsigned char *sha2,
 670                                    enum object_type type1,
 671                                    enum object_type type2)
 672 {
 673         int cmp = type1 - type2;
 674         if (cmp)
 675                 return cmp;
 676         return hashcmp(sha1, sha2);
 677 }
 678
 679 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
 680 {
 681         int first = 0, last = nr_ref_deltas;
 682
 683         while (first < last) {
 684                 int next = (first + last) / 2;
 685                 struct ref_delta_entry *delta = &ref_deltas[next];
 686                 int cmp;
 687
 688                 cmp = compare_ref_delta_bases(sha1, delta->sha1,
 689                                               type, objects[delta->obj_no].type);
 690                 if (!cmp)
 691                         return next;
 692                 if (cmp < 0) {
 693                         last = next;
 694                         continue;
 695                 }
 696                 first = next+1;
 697         }
 698         return -first-1;
 699 }
 700
 701 static void find_ref_delta_children(const unsigned char *sha1,
 702                                     int *first_index, int *last_index,
 703                                     enum object_type type)
 704 {
 705         int first = find_ref_delta(sha1, type);
 706         int last = first;
 707         int end = nr_ref_deltas - 1;
 708
 709         if (first < 0) {
 710                 *first_index = 0;
 711                 *last_index = -1;
 712                 return;
 713         }
 714         while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
 715                 --first;
 716         while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
 717                 ++last;
 718         *first_index = first;
 719         *last_index = last;
 720 }
 721
 722 struct compare_data {
 723         struct object_entry *entry;
 724         struct git_istream *st;
 725         unsigned char *buf;
 726         unsigned long buf_size;
 727 };
 728
 729 static int compare_objects(const unsigned char *buf, unsigned long size,
 730                            void *cb_data)
 731 {
 732         struct compare_data *data = cb_data;
 733
 734         if (data->buf_size < size) {
 735                 free(data->buf);
 736                 data->buf = xmalloc(size);
 737                 data->buf_size = size;
 738         }
 739
 740         while (size) {
 741                 ssize_t len = read_istream(data->st, data->buf, size);
 742                 if (len == 0)
 743                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 744                             sha1_to_hex(data->entry->idx.sha1));
 745                 if (len < 0)
 746                         die(_("unable to read %s"),
 747                             sha1_to_hex(data->entry->idx.sha1));
 748                 if (memcmp(buf, data->buf, len))
 749                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 750                             sha1_to_hex(data->entry->idx.sha1));
 751                 size -= len;
 752                 buf += len;
 753         }
 754         return 0;
 755 }
 756
 757 static int check_collison(struct object_entry *entry)
 758 {
 759         struct compare_data data;
 760         enum object_type type;
 761         unsigned long size;
 762
 763         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 764                 return -1;
 765
 766         memset(&data, 0, sizeof(data));
 767         data.entry = entry;
 768         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 769         if (!data.st)
 770                 return -1;
 771         if (size != entry->size || type != entry->type)
 772                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 773                     sha1_to_hex(entry->idx.sha1));
 774         unpack_data(entry, compare_objects, &data);
 775         close_istream(data.st);
 776         free(data.buf);
 777         return 0;
 778 }
 779
 780 static void sha1_object(const void *data, struct object_entry *obj_entry,
 781                         unsigned long size, enum object_type type,
 782                         const unsigned char *sha1)
 783 {
 784         void *new_data = NULL;
 785         int collision_test_needed;
 786
 787         assert(data || obj_entry);
 788
 789         read_lock();
 790         collision_test_needed = has_sha1_file_with_flags(sha1, HAS_SHA1_QUICK);
 791         read_unlock();
 792
 793         if (collision_test_needed && !data) {
 794                 read_lock();
 795                 if (!check_collison(obj_entry))
 796                         collision_test_needed = 0;
 797                 read_unlock();
 798         }
 799         if (collision_test_needed) {
 800                 void *has_data;
 801                 enum object_type has_type;
 802                 unsigned long has_size;
 803                 read_lock();
 804                 has_type = sha1_object_info(sha1, &has_size);
 805                 if (has_type != type || has_size != size)
 806                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 807                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 808                 read_unlock();
 809                 if (!data)
 810                         data = new_data = get_data_from_pack(obj_entry);
 811                 if (!has_data)
 812                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 813                 if (size != has_size || type != has_type ||
 814                     memcmp(data, has_data, size) != 0)
 815                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 816                 free(has_data);
 817         }
 818
 819         if (strict) {
 820                 read_lock();
 821                 if (type == OBJ_BLOB) {
 822                         struct blob *blob = lookup_blob(sha1);
 823                         if (blob)
 824                                 blob->object.flags |= FLAG_CHECKED;
 825                         else
 826                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 827                 } else {
 828                         struct object *obj;
 829                         int eaten;
 830                         void *buf = (void *) data;
 831
 832                         assert(data && "data can only be NULL for large _blobs_");
 833
 834                         /*
 835                          * we do not need to free the memory here, as the
 836                          * buf is deleted by the caller.
 837                          */
 838                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 839                         if (!obj)
 840                                 die(_("invalid %s"), typename(type));
 841                         if (do_fsck_object &&
 842                             fsck_object(obj, buf, size, &fsck_options))
 843                                 die(_("Error in object"));
 844                         if (fsck_walk(obj, NULL, &fsck_options))
 845                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 846
 847                         if (obj->type == OBJ_TREE) {
 848                                 struct tree *item = (struct tree *) obj;
 849                                 item->buffer = NULL;
 850                                 obj->parsed = 0;
 851                         }
 852                         if (obj->type == OBJ_COMMIT) {
 853                                 struct commit *commit = (struct commit *) obj;
 854                                 if (detach_commit_buffer(commit, NULL) != data)
 855                                         die("BUG: parse_object_buffer transmogrified our buffer");
 856                         }
 857                         obj->flags |= FLAG_CHECKED;
 858                 }
 859                 read_unlock();
 860         }
 861
 862         free(new_data);
 863 }
 864
 865 /*
 866  * This function is part of find_unresolved_deltas(). There are two
 867  * walkers going in the opposite ways.
 868  *
 869  * The first one in find_unresolved_deltas() traverses down from
 870  * parent node to children, deflating nodes along the way. However,
 871  * memory for deflated nodes is limited by delta_base_cache_limit, so
 872  * at some point parent node's deflated content may be freed.
 873  *
 874  * The second walker is this function, which goes from current node up
 875  * to top parent if necessary to deflate the node. In normal
 876  * situation, its parent node would be already deflated, so it just
 877  * needs to apply delta.
 878  *
 879  * In the worst case scenario, parent node is no longer deflated because
 880  * we're running out of delta_base_cache_limit; we need to re-deflate
 881  * parents, possibly up to the top base.
 882  *
 883  * All deflated objects here are subject to be freed if we exceed
 884  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 885  * just need to make sure the last node is not freed.
 886  */
 887 static void *get_base_data(struct base_data *c)
 888 {
 889         if (!c->data) {
 890                 struct object_entry *obj = c->obj;
 891                 struct base_data **delta = NULL;
 892                 int delta_nr = 0, delta_alloc = 0;
 893
 894                 while (is_delta_type(c->obj->type) && !c->data) {
 895                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 896                         delta[delta_nr++] = c;
 897                         c = c->base;
 898                 }
 899                 if (!delta_nr) {
 900                         c->data = get_data_from_pack(obj);
 901                         c->size = obj->size;
 902                         get_thread_data()->base_cache_used += c->size;
 903                         prune_base_data(c);
 904                 }
 905                 for (; delta_nr > 0; delta_nr--) {
 906                         void *base, *raw;
 907                         c = delta[delta_nr - 1];
 908                         obj = c->obj;
 909                         base = get_base_data(c->base);
 910                         raw = get_data_from_pack(obj);
 911                         c->data = patch_delta(
 912                                 base, c->base->size,
 913                                 raw, obj->size,
 914                                 &c->size);
 915                         free(raw);
 916                         if (!c->data)
 917                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 918                         get_thread_data()->base_cache_used += c->size;
 919                         prune_base_data(c);
 920                 }
 921                 free(delta);
 922         }
 923         return c->data;
 924 }
 925
 926 static void resolve_delta(struct object_entry *delta_obj,
 927                           struct base_data *base, struct base_data *result)
 928 {
 929         void *base_data, *delta_data;
 930
 931         if (show_stat) {
 932                 int i = delta_obj - objects;
 933                 int j = base->obj - objects;
 934                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 935                 deepest_delta_lock();
 936                 if (deepest_delta < obj_stat[i].delta_depth)
 937                         deepest_delta = obj_stat[i].delta_depth;
 938                 deepest_delta_unlock();
 939                 obj_stat[i].base_object_no = j;
 940         }
 941         delta_data = get_data_from_pack(delta_obj);
 942         base_data = get_base_data(base);
 943         result->obj = delta_obj;
 944         result->data = patch_delta(base_data, base->size,
 945                                    delta_data, delta_obj->size, &result->size);
 946         free(delta_data);
 947         if (!result->data)
 948                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 949         hash_sha1_file(result->data, result->size,
 950                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 951         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 952                     delta_obj->idx.sha1);
 953         counter_lock();
 954         nr_resolved_deltas++;
 955         counter_unlock();
 956 }
 957
 958 /*
 959  * Standard boolean compare-and-swap: atomically check whether "*type" is
 960  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 961  * and return false.
 962  */
 963 static int compare_and_swap_type(signed char *type,
 964                                  enum object_type want,
 965                                  enum object_type set)
 966 {
 967         enum object_type old;
 968
 969         type_cas_lock();
 970         old = *type;
 971         if (old == want)
 972                 *type = set;
 973         type_cas_unlock();
 974
 975         return old == want;
 976 }
 977
 978 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 979                                                   struct base_data *prev_base)
 980 {
 981         if (base->ref_last == -1 && base->ofs_last == -1) {
 982                 find_ref_delta_children(base->obj->idx.sha1,
 983                                         &base->ref_first, &base->ref_last,
 984                                         OBJ_REF_DELTA);
 985
 986                 find_ofs_delta_children(base->obj->idx.offset,
 987                                         &base->ofs_first, &base->ofs_last,
 988                                         OBJ_OFS_DELTA);
 989
 990                 if (base->ref_last == -1 && base->ofs_last == -1) {
 991                         free(base->data);
 992                         return NULL;
 993                 }
 994
 995                 link_base_data(prev_base, base);
 996         }
 997
 998         if (base->ref_first <= base->ref_last) {
 999                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1000                 struct base_data *result = alloc_base_data();
1001
1002                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1003                                            base->obj->real_type))
1004                         die("BUG: child->real_type != OBJ_REF_DELTA");
1005
1006                 resolve_delta(child, base, result);
1007                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1008                         free_base_data(base);
1009
1010                 base->ref_first++;
1011                 return result;
1012         }
1013
1014         if (base->ofs_first <= base->ofs_last) {
1015                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1016                 struct base_data *result = alloc_base_data();
1017
1018                 assert(child->real_type == OBJ_OFS_DELTA);
1019                 child->real_type = base->obj->real_type;
1020                 resolve_delta(child, base, result);
1021                 if (base->ofs_first == base->ofs_last)
1022                         free_base_data(base);
1023
1024                 base->ofs_first++;
1025                 return result;
1026         }
1027
1028         unlink_base_data(base);
1029         return NULL;
1030 }
1031
1032 static void find_unresolved_deltas(struct base_data *base)
1033 {
1034         struct base_data *new_base, *prev_base = NULL;
1035         for (;;) {
1036                 new_base = find_unresolved_deltas_1(base, prev_base);
1037
1038                 if (new_base) {
1039                         prev_base = base;
1040                         base = new_base;
1041                 } else {
1042                         free(base);
1043                         base = prev_base;
1044                         if (!base)
1045                                 return;
1046                         prev_base = base->base;
1047                 }
1048         }
1049 }
1050
1051 static int compare_ofs_delta_entry(const void *a, const void *b)
1052 {
1053         const struct ofs_delta_entry *delta_a = a;
1054         const struct ofs_delta_entry *delta_b = b;
1055
1056         return delta_a->offset < delta_b->offset ? -1 :
1057                delta_a->offset > delta_b->offset ?  1 :
1058                0;
1059 }
1060
1061 static int compare_ref_delta_entry(const void *a, const void *b)
1062 {
1063         const struct ref_delta_entry *delta_a = a;
1064         const struct ref_delta_entry *delta_b = b;
1065
1066         return hashcmp(delta_a->sha1, delta_b->sha1);
1067 }
1068
1069 static void resolve_base(struct object_entry *obj)
1070 {
1071         struct base_data *base_obj = alloc_base_data();
1072         base_obj->obj = obj;
1073         base_obj->data = NULL;
1074         find_unresolved_deltas(base_obj);
1075 }
1076
1077 #ifndef NO_PTHREADS
1078 static void *threaded_second_pass(void *data)
1079 {
1080         set_thread_data(data);
1081         for (;;) {
1082                 int i;
1083                 counter_lock();
1084                 display_progress(progress, nr_resolved_deltas);
1085                 counter_unlock();
1086                 work_lock();
1087                 while (nr_dispatched < nr_objects &&
1088                        is_delta_type(objects[nr_dispatched].type))
1089                         nr_dispatched++;
1090                 if (nr_dispatched >= nr_objects) {
1091                         work_unlock();
1092                         break;
1093                 }
1094                 i = nr_dispatched++;
1095                 work_unlock();
1096
1097                 resolve_base(&objects[i]);
1098         }
1099         return NULL;
1100 }
1101 #endif
1102
1103 /*
1104  * First pass:
1105  * - find locations of all objects;
1106  * - calculate SHA1 of all non-delta objects;
1107  * - remember base (SHA1 or offset) for all deltas.
1108  */
1109 static void parse_pack_objects(unsigned char *sha1)
1110 {
1111         int i, nr_delays = 0;
1112         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1113         unsigned char ref_delta_sha1[20];
1114         struct stat st;
1115
1116         if (verbose)
1117                 progress = start_progress(
1118                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1119                                 nr_objects);
1120         for (i = 0; i < nr_objects; i++) {
1121                 struct object_entry *obj = &objects[i];
1122                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1123                                               ref_delta_sha1, obj->idx.sha1);
1124                 obj->real_type = obj->type;
1125                 if (obj->type == OBJ_OFS_DELTA) {
1126                         nr_ofs_deltas++;
1127                         ofs_delta->obj_no = i;
1128                         ofs_delta++;
1129                 } else if (obj->type == OBJ_REF_DELTA) {
1130                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1131                         hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1);
1132                         ref_deltas[nr_ref_deltas].obj_no = i;
1133                         nr_ref_deltas++;
1134                 } else if (!data) {
1135                         /* large blobs, check later */
1136                         obj->real_type = OBJ_BAD;
1137                         nr_delays++;
1138                 } else
1139                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1140                 free(data);
1141                 display_progress(progress, i+1);
1142         }
1143         objects[i].idx.offset = consumed_bytes;
1144         stop_progress(&progress);
1145
1146         /* Check pack integrity */
1147         flush();
1148         git_SHA1_Final(sha1, &input_ctx);
1149         if (hashcmp(fill(20), sha1))
1150                 die(_("pack is corrupted (SHA1 mismatch)"));
1151         use(20);
1152
1153         /* If input_fd is a file, we should have reached its end now. */
1154         if (fstat(input_fd, &st))
1155                 die_errno(_("cannot fstat packfile"));
1156         if (S_ISREG(st.st_mode) &&
1157                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1158                 die(_("pack has junk at the end"));
1159
1160         for (i = 0; i < nr_objects; i++) {
1161                 struct object_entry *obj = &objects[i];
1162                 if (obj->real_type != OBJ_BAD)
1163                         continue;
1164                 obj->real_type = obj->type;
1165                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1166                 nr_delays--;
1167         }
1168         if (nr_delays)
1169                 die(_("confusion beyond insanity in parse_pack_objects()"));
1170 }
1171
1172 /*
1173  * Second pass:
1174  * - for all non-delta objects, look if it is used as a base for
1175  *   deltas;
1176  * - if used as a base, uncompress the object and apply all deltas,
1177  *   recursively checking if the resulting object is used as a base
1178  *   for some more deltas.
1179  */
1180 static void resolve_deltas(void)
1181 {
1182         int i;
1183
1184         if (!nr_ofs_deltas && !nr_ref_deltas)
1185                 return;
1186
1187         /* Sort deltas by base SHA1/offset for fast searching */
1188         qsort(ofs_deltas, nr_ofs_deltas, sizeof(struct ofs_delta_entry),
1189               compare_ofs_delta_entry);
1190         qsort(ref_deltas, nr_ref_deltas, sizeof(struct ref_delta_entry),
1191               compare_ref_delta_entry);
1192
1193         if (verbose)
1194                 progress = start_progress(_("Resolving deltas"),
1195                                           nr_ref_deltas + nr_ofs_deltas);
1196
1197 #ifndef NO_PTHREADS
1198         nr_dispatched = 0;
1199         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1200                 init_thread();
1201                 for (i = 0; i < nr_threads; i++) {
1202                         int ret = pthread_create(&thread_data[i].thread, NULL,
1203                                                  threaded_second_pass, thread_data + i);
1204                         if (ret)
1205                                 die(_("unable to create thread: %s"),
1206                                     strerror(ret));
1207                 }
1208                 for (i = 0; i < nr_threads; i++)
1209                         pthread_join(thread_data[i].thread, NULL);
1210                 cleanup_thread();
1211                 return;
1212         }
1213 #endif
1214
1215         for (i = 0; i < nr_objects; i++) {
1216                 struct object_entry *obj = &objects[i];
1217
1218                 if (is_delta_type(obj->type))
1219                         continue;
1220                 resolve_base(obj);
1221                 display_progress(progress, nr_resolved_deltas);
1222         }
1223 }
1224
1225 /*
1226  * Third pass:
1227  * - append objects to convert thin pack to full pack if required
1228  * - write the final 20-byte SHA-1
1229  */
1230 static void fix_unresolved_deltas(struct sha1file *f);
1231 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1232 {
1233         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1234                 stop_progress(&progress);
1235                 /* Flush remaining pack final 20-byte SHA1. */
1236                 flush();
1237                 return;
1238         }
1239
1240         if (fix_thin_pack) {
1241                 struct sha1file *f;
1242                 unsigned char read_sha1[20], tail_sha1[20];
1243                 struct strbuf msg = STRBUF_INIT;
1244                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1245                 int nr_objects_initial = nr_objects;
1246                 if (nr_unresolved <= 0)
1247                         die(_("confusion beyond insanity"));
1248                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1249                 memset(objects + nr_objects + 1, 0,
1250                        nr_unresolved * sizeof(*objects));
1251                 f = sha1fd(output_fd, curr_pack);
1252                 fix_unresolved_deltas(f);
1253                 strbuf_addf(&msg, _("completed with %d local objects"),
1254                             nr_objects - nr_objects_initial);
1255                 stop_progress_msg(&progress, msg.buf);
1256                 strbuf_release(&msg);
1257                 sha1close(f, tail_sha1, 0);
1258                 hashcpy(read_sha1, pack_sha1);
1259                 fixup_pack_header_footer(output_fd, pack_sha1,
1260                                          curr_pack, nr_objects,
1261                                          read_sha1, consumed_bytes-20);
1262                 if (hashcmp(read_sha1, tail_sha1) != 0)
1263                         die(_("Unexpected tail checksum for %s "
1264                               "(disk corruption?)"), curr_pack);
1265         }
1266         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1267                 die(Q_("pack has %d unresolved delta",
1268                        "pack has %d unresolved deltas",
1269                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1270                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1271 }
1272
1273 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1274 {
1275         git_zstream stream;
1276         int status;
1277         unsigned char outbuf[4096];
1278
1279         git_deflate_init(&stream, zlib_compression_level);
1280         stream.next_in = in;
1281         stream.avail_in = size;
1282
1283         do {
1284                 stream.next_out = outbuf;
1285                 stream.avail_out = sizeof(outbuf);
1286                 status = git_deflate(&stream, Z_FINISH);
1287                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1288         } while (status == Z_OK);
1289
1290         if (status != Z_STREAM_END)
1291                 die(_("unable to deflate appended object (%d)"), status);
1292         size = stream.total_out;
1293         git_deflate_end(&stream);
1294         return size;
1295 }
1296
1297 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1298                                const unsigned char *sha1, void *buf,
1299                                unsigned long size, enum object_type type)
1300 {
1301         struct object_entry *obj = &objects[nr_objects++];
1302         unsigned char header[10];
1303         unsigned long s = size;
1304         int n = 0;
1305         unsigned char c = (type << 4) | (s & 15);
1306         s >>= 4;
1307         while (s) {
1308                 header[n++] = c | 0x80;
1309                 c = s & 0x7f;
1310                 s >>= 7;
1311         }
1312         header[n++] = c;
1313         crc32_begin(f);
1314         sha1write(f, header, n);
1315         obj[0].size = size;
1316         obj[0].hdr_size = n;
1317         obj[0].type = type;
1318         obj[0].real_type = type;
1319         obj[1].idx.offset = obj[0].idx.offset + n;
1320         obj[1].idx.offset += write_compressed(f, buf, size);
1321         obj[0].idx.crc32 = crc32_end(f);
1322         sha1flush(f);
1323         hashcpy(obj->idx.sha1, sha1);
1324         return obj;
1325 }
1326
1327 static int delta_pos_compare(const void *_a, const void *_b)
1328 {
1329         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1330         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1331         return a->obj_no - b->obj_no;
1332 }
1333
1334 static void fix_unresolved_deltas(struct sha1file *f)
1335 {
1336         struct ref_delta_entry **sorted_by_pos;
1337         int i;
1338
1339         /*
1340          * Since many unresolved deltas may well be themselves base objects
1341          * for more unresolved deltas, we really want to include the
1342          * smallest number of base objects that would cover as much delta
1343          * as possible by picking the
1344          * trunc deltas first, allowing for other deltas to resolve without
1345          * additional base objects.  Since most base objects are to be found
1346          * before deltas depending on them, a good heuristic is to start
1347          * resolving deltas in the same order as their position in the pack.
1348          */
1349         sorted_by_pos = xmalloc(nr_ref_deltas * sizeof(*sorted_by_pos));
1350         for (i = 0; i < nr_ref_deltas; i++)
1351                 sorted_by_pos[i] = &ref_deltas[i];
1352         qsort(sorted_by_pos, nr_ref_deltas, sizeof(*sorted_by_pos), delta_pos_compare);
1353
1354         for (i = 0; i < nr_ref_deltas; i++) {
1355                 struct ref_delta_entry *d = sorted_by_pos[i];
1356                 enum object_type type;
1357                 struct base_data *base_obj = alloc_base_data();
1358
1359                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1360                         continue;
1361                 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1362                 if (!base_obj->data)
1363                         continue;
1364
1365                 if (check_sha1_signature(d->sha1, base_obj->data,
1366                                 base_obj->size, typename(type)))
1367                         die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1368                 base_obj->obj = append_obj_to_pack(f, d->sha1,
1369                                         base_obj->data, base_obj->size, type);
1370                 find_unresolved_deltas(base_obj);
1371                 display_progress(progress, nr_resolved_deltas);
1372         }
1373         free(sorted_by_pos);
1374 }
1375
1376 static void final(const char *final_pack_name, const char *curr_pack_name,
1377                   const char *final_index_name, const char *curr_index_name,
1378                   const char *keep_name, const char *keep_msg,
1379                   unsigned char *sha1)
1380 {
1381         const char *report = "pack";
1382         char name[PATH_MAX];
1383         int err;
1384
1385         if (!from_stdin) {
1386                 close(input_fd);
1387         } else {
1388                 fsync_or_die(output_fd, curr_pack_name);
1389                 err = close(output_fd);
1390                 if (err)
1391                         die_errno(_("error while closing pack file"));
1392         }
1393
1394         if (keep_msg) {
1395                 int keep_fd, keep_msg_len = strlen(keep_msg);
1396
1397                 if (!keep_name)
1398                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1399                 else
1400                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1401
1402                 if (keep_fd < 0) {
1403                         if (errno != EEXIST)
1404                                 die_errno(_("cannot write keep file '%s'"),
1405                                           keep_name ? keep_name : name);
1406                 } else {
1407                         if (keep_msg_len > 0) {
1408                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1409                                 write_or_die(keep_fd, "\n", 1);
1410                         }
1411                         if (close(keep_fd) != 0)
1412                                 die_errno(_("cannot close written keep file '%s'"),
1413                                           keep_name ? keep_name : name);
1414                         report = "keep";
1415                 }
1416         }
1417
1418         if (final_pack_name != curr_pack_name) {
1419                 if (!final_pack_name) {
1420                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1421                                  get_object_directory(), sha1_to_hex(sha1));
1422                         final_pack_name = name;
1423                 }
1424                 if (finalize_object_file(curr_pack_name, final_pack_name))
1425                         die(_("cannot store pack file"));
1426         } else if (from_stdin)
1427                 chmod(final_pack_name, 0444);
1428
1429         if (final_index_name != curr_index_name) {
1430                 if (!final_index_name) {
1431                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1432                                  get_object_directory(), sha1_to_hex(sha1));
1433                         final_index_name = name;
1434                 }
1435                 if (finalize_object_file(curr_index_name, final_index_name))
1436                         die(_("cannot store index file"));
1437         } else
1438                 chmod(final_index_name, 0444);
1439
1440         if (!from_stdin) {
1441                 printf("%s\n", sha1_to_hex(sha1));
1442         } else {
1443                 char buf[48];
1444                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1445                                    report, sha1_to_hex(sha1));
1446                 write_or_die(1, buf, len);
1447
1448                 /*
1449                  * Let's just mimic git-unpack-objects here and write
1450                  * the last part of the input buffer to stdout.
1451                  */
1452                 while (input_len) {
1453                         err = xwrite(1, input_buffer + input_offset, input_len);
1454                         if (err <= 0)
1455                                 break;
1456                         input_len -= err;
1457                         input_offset += err;
1458                 }
1459         }
1460 }
1461
1462 static int git_index_pack_config(const char *k, const char *v, void *cb)
1463 {
1464         struct pack_idx_option *opts = cb;
1465
1466         if (!strcmp(k, "pack.indexversion")) {
1467                 opts->version = git_config_int(k, v);
1468                 if (opts->version > 2)
1469                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1470                 return 0;
1471         }
1472         if (!strcmp(k, "pack.threads")) {
1473                 nr_threads = git_config_int(k, v);
1474                 if (nr_threads < 0)
1475                         die(_("invalid number of threads specified (%d)"),
1476                             nr_threads);
1477 #ifdef NO_PTHREADS
1478                 if (nr_threads != 1)
1479                         warning(_("no threads support, ignoring %s"), k);
1480                 nr_threads = 1;
1481 #endif
1482                 return 0;
1483         }
1484         return git_default_config(k, v, cb);
1485 }
1486
1487 static int cmp_uint32(const void *a_, const void *b_)
1488 {
1489         uint32_t a = *((uint32_t *)a_);
1490         uint32_t b = *((uint32_t *)b_);
1491
1492         return (a < b) ? -1 : (a != b);
1493 }
1494
1495 static void read_v2_anomalous_offsets(struct packed_git *p,
1496                                       struct pack_idx_option *opts)
1497 {
1498         const uint32_t *idx1, *idx2;
1499         uint32_t i;
1500
1501         /* The address of the 4-byte offset table */
1502         idx1 = (((const uint32_t *)p->index_data)
1503                 + 2 /* 8-byte header */
1504                 + 256 /* fan out */
1505                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1506                 + p->num_objects /* CRC32 table */
1507                 );
1508
1509         /* The address of the 8-byte offset table */
1510         idx2 = idx1 + p->num_objects;
1511
1512         for (i = 0; i < p->num_objects; i++) {
1513                 uint32_t off = ntohl(idx1[i]);
1514                 if (!(off & 0x80000000))
1515                         continue;
1516                 off = off & 0x7fffffff;
1517                 if (idx2[off * 2])
1518                         continue;
1519                 /*
1520                  * The real offset is ntohl(idx2[off * 2]) in high 4
1521                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1522                  * octets.  But idx2[off * 2] is Zero!!!
1523                  */
1524                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1525                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1526         }
1527
1528         if (1 < opts->anomaly_nr)
1529                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1530 }
1531
1532 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1533 {
1534         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1535
1536         if (!p)
1537                 die(_("Cannot open existing pack file '%s'"), pack_name);
1538         if (open_pack_index(p))
1539                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1540
1541         /* Read the attributes from the existing idx file */
1542         opts->version = p->index_version;
1543
1544         if (opts->version == 2)
1545                 read_v2_anomalous_offsets(p, opts);
1546
1547         /*
1548          * Get rid of the idx file as we do not need it anymore.
1549          * NEEDSWORK: extract this bit from free_pack_by_name() in
1550          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1551          * know we haven't installed this pack (hence we never have
1552          * read anything from it).
1553          */
1554         close_pack_index(p);
1555         free(p);
1556 }
1557
1558 static void show_pack_info(int stat_only)
1559 {
1560         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1561         unsigned long *chain_histogram = NULL;
1562
1563         if (deepest_delta)
1564                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1565
1566         for (i = 0; i < nr_objects; i++) {
1567                 struct object_entry *obj = &objects[i];
1568
1569                 if (is_delta_type(obj->type))
1570                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1571                 if (stat_only)
1572                         continue;
1573                 printf("%s %-6s %lu %lu %"PRIuMAX,
1574                        sha1_to_hex(obj->idx.sha1),
1575                        typename(obj->real_type), obj->size,
1576                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1577                        (uintmax_t)obj->idx.offset);
1578                 if (is_delta_type(obj->type)) {
1579                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1580                         printf(" %u %s", obj_stat[i].delta_depth, sha1_to_hex(bobj->idx.sha1));
1581                 }
1582                 putchar('\n');
1583         }
1584
1585         if (baseobjects)
1586                 printf_ln(Q_("non delta: %d object",
1587                              "non delta: %d objects",
1588                              baseobjects),
1589                           baseobjects);
1590         for (i = 0; i < deepest_delta; i++) {
1591                 if (!chain_histogram[i])
1592                         continue;
1593                 printf_ln(Q_("chain length = %d: %lu object",
1594                              "chain length = %d: %lu objects",
1595                              chain_histogram[i]),
1596                           i + 1,
1597                           chain_histogram[i]);
1598         }
1599 }
1600
1601 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1602 {
1603         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1604         const char *curr_index;
1605         const char *index_name = NULL, *pack_name = NULL;
1606         const char *keep_name = NULL, *keep_msg = NULL;
1607         struct strbuf index_name_buf = STRBUF_INIT,
1608                       keep_name_buf = STRBUF_INIT;
1609         struct pack_idx_entry **idx_objects;
1610         struct pack_idx_option opts;
1611         unsigned char pack_sha1[20];
1612         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1613
1614         if (argc == 2 && !strcmp(argv[1], "-h"))
1615                 usage(index_pack_usage);
1616
1617         check_replace_refs = 0;
1618         fsck_options.walk = mark_link;
1619
1620         reset_pack_idx_option(&opts);
1621         git_config(git_index_pack_config, &opts);
1622         if (prefix && chdir(prefix))
1623                 die(_("Cannot come back to cwd"));
1624
1625         for (i = 1; i < argc; i++) {
1626                 const char *arg = argv[i];
1627
1628                 if (*arg == '-') {
1629                         if (!strcmp(arg, "--stdin")) {
1630                                 from_stdin = 1;
1631                         } else if (!strcmp(arg, "--fix-thin")) {
1632                                 fix_thin_pack = 1;
1633                         } else if (!strcmp(arg, "--strict")) {
1634                                 strict = 1;
1635                                 do_fsck_object = 1;
1636                         } else if (skip_prefix(arg, "--strict=", &arg)) {
1637                                 strict = 1;
1638                                 do_fsck_object = 1;
1639                                 fsck_set_msg_types(&fsck_options, arg);
1640                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1641                                 strict = 1;
1642                                 check_self_contained_and_connected = 1;
1643                         } else if (!strcmp(arg, "--verify")) {
1644                                 verify = 1;
1645                         } else if (!strcmp(arg, "--verify-stat")) {
1646                                 verify = 1;
1647                                 show_stat = 1;
1648                         } else if (!strcmp(arg, "--verify-stat-only")) {
1649                                 verify = 1;
1650                                 show_stat = 1;
1651                                 stat_only = 1;
1652                         } else if (!strcmp(arg, "--keep")) {
1653                                 keep_msg = "";
1654                         } else if (starts_with(arg, "--keep=")) {
1655                                 keep_msg = arg + 7;
1656                         } else if (starts_with(arg, "--threads=")) {
1657                                 char *end;
1658                                 nr_threads = strtoul(arg+10, &end, 0);
1659                                 if (!arg[10] || *end || nr_threads < 0)
1660                                         usage(index_pack_usage);
1661 #ifdef NO_PTHREADS
1662                                 if (nr_threads != 1)
1663                                         warning(_("no threads support, "
1664                                                   "ignoring %s"), arg);
1665                                 nr_threads = 1;
1666 #endif
1667                         } else if (starts_with(arg, "--pack_header=")) {
1668                                 struct pack_header *hdr;
1669                                 char *c;
1670
1671                                 hdr = (struct pack_header *)input_buffer;
1672                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1673                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1674                                 if (*c != ',')
1675                                         die(_("bad %s"), arg);
1676                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1677                                 if (*c)
1678                                         die(_("bad %s"), arg);
1679                                 input_len = sizeof(*hdr);
1680                         } else if (!strcmp(arg, "-v")) {
1681                                 verbose = 1;
1682                         } else if (!strcmp(arg, "-o")) {
1683                                 if (index_name || (i+1) >= argc)
1684                                         usage(index_pack_usage);
1685                                 index_name = argv[++i];
1686                         } else if (starts_with(arg, "--index-version=")) {
1687                                 char *c;
1688                                 opts.version = strtoul(arg + 16, &c, 10);
1689                                 if (opts.version > 2)
1690                                         die(_("bad %s"), arg);
1691                                 if (*c == ',')
1692                                         opts.off32_limit = strtoul(c+1, &c, 0);
1693                                 if (*c || opts.off32_limit & 0x80000000)
1694                                         die(_("bad %s"), arg);
1695                         } else
1696                                 usage(index_pack_usage);
1697                         continue;
1698                 }
1699
1700                 if (pack_name)
1701                         usage(index_pack_usage);
1702                 pack_name = arg;
1703         }
1704
1705         if (!pack_name && !from_stdin)
1706                 usage(index_pack_usage);
1707         if (fix_thin_pack && !from_stdin)
1708                 die(_("--fix-thin cannot be used without --stdin"));
1709         if (!index_name && pack_name) {
1710                 size_t len;
1711                 if (!strip_suffix(pack_name, ".pack", &len))
1712                         die(_("packfile name '%s' does not end with '.pack'"),
1713                             pack_name);
1714                 strbuf_add(&index_name_buf, pack_name, len);
1715                 strbuf_addstr(&index_name_buf, ".idx");
1716                 index_name = index_name_buf.buf;
1717         }
1718         if (keep_msg && !keep_name && pack_name) {
1719                 size_t len;
1720                 if (!strip_suffix(pack_name, ".pack", &len))
1721                         die(_("packfile name '%s' does not end with '.pack'"),
1722                             pack_name);
1723                 strbuf_add(&keep_name_buf, pack_name, len);
1724                 strbuf_addstr(&keep_name_buf, ".idx");
1725                 keep_name = keep_name_buf.buf;
1726         }
1727         if (verify) {
1728                 if (!index_name)
1729                         die(_("--verify with no packfile name given"));
1730                 read_idx_option(&opts, index_name);
1731                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1732         }
1733         if (strict)
1734                 opts.flags |= WRITE_IDX_STRICT;
1735
1736 #ifndef NO_PTHREADS
1737         if (!nr_threads) {
1738                 nr_threads = online_cpus();
1739                 /* An experiment showed that more threads does not mean faster */
1740                 if (nr_threads > 3)
1741                         nr_threads = 3;
1742         }
1743 #endif
1744
1745         curr_pack = open_pack_file(pack_name);
1746         parse_pack_header();
1747         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1748         if (show_stat)
1749                 obj_stat = xcalloc(nr_objects + 1, sizeof(struct object_stat));
1750         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1751         parse_pack_objects(pack_sha1);
1752         resolve_deltas();
1753         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1754         free(ofs_deltas);
1755         free(ref_deltas);
1756         if (strict)
1757                 foreign_nr = check_objects();
1758
1759         if (show_stat)
1760                 show_pack_info(stat_only);
1761
1762         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1763         for (i = 0; i < nr_objects; i++)
1764                 idx_objects[i] = &objects[i].idx;
1765         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1766         free(idx_objects);
1767
1768         if (!verify)
1769                 final(pack_name, curr_pack,
1770                       index_name, curr_index,
1771                       keep_name, keep_msg,
1772                       pack_sha1);
1773         else
1774                 close(input_fd);
1775         free(objects);
1776         strbuf_release(&index_name_buf);
1777         strbuf_release(&keep_name_buf);
1778         if (pack_name == NULL)
1779                 free((void *) curr_pack);
1780         if (index_name == NULL)
1781                 free((void *) curr_index);
1782
1783         /*
1784          * Let the caller know this pack is not self contained
1785          */
1786         if (check_self_contained_and_connected && foreign_nr)
1787                 return 1;
1788
1789         return 0;
1790 }