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

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