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 struct thread_local {
  44 #ifndef NO_PTHREADS
  45         pthread_t thread;
  46 #endif
  47         struct base_data *base_cache;
  48         size_t base_cache_used;
  49         int pack_fd;
  50 };
  51
  52 /*
  53  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  54  * to memcmp() only the first 20 bytes.
  55  */
  56 #define UNION_BASE_SZ   20
  57
  58 #define FLAG_LINK (1u<<20)
  59 #define FLAG_CHECKED (1u<<21)
  60
  61 struct delta_entry {
  62         union delta_base base;
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct delta_entry *deltas;
  68 static struct thread_local nothread_data;
  69 static int nr_objects;
  70 static int nr_deltas;
  71 static int nr_resolved_deltas;
  72 static int nr_threads;
  73
  74 static int from_stdin;
  75 static int strict;
  76 static int do_fsck_object;
  77 static int verbose;
  78 static int show_stat;
  79 static int check_self_contained_and_connected;
  80
  81 static struct progress *progress;
  82
  83 /* We always read in 4kB chunks. */
  84 static unsigned char input_buffer[4096];
  85 static unsigned int input_offset, input_len;
  86 static off_t consumed_bytes;
  87 static unsigned deepest_delta;
  88 static git_SHA_CTX input_ctx;
  89 static uint32_t input_crc32;
  90 static int input_fd, output_fd;
  91 static const char *curr_pack;
  92
  93 #ifndef NO_PTHREADS
  94
  95 static struct thread_local *thread_data;
  96 static int nr_dispatched;
  97 static int threads_active;
  98
  99 static pthread_mutex_t read_mutex;
 100 #define read_lock()             lock_mutex(&read_mutex)
 101 #define read_unlock()           unlock_mutex(&read_mutex)
 102
 103 static pthread_mutex_t counter_mutex;
 104 #define counter_lock()          lock_mutex(&counter_mutex)
 105 #define counter_unlock()        unlock_mutex(&counter_mutex)
 106
 107 static pthread_mutex_t work_mutex;
 108 #define work_lock()             lock_mutex(&work_mutex)
 109 #define work_unlock()           unlock_mutex(&work_mutex)
 110
 111 static pthread_mutex_t deepest_delta_mutex;
 112 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 113 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 114
 115 static pthread_key_t key;
 116
 117 static inline void lock_mutex(pthread_mutex_t *mutex)
 118 {
 119         if (threads_active)
 120                 pthread_mutex_lock(mutex);
 121 }
 122
 123 static inline void unlock_mutex(pthread_mutex_t *mutex)
 124 {
 125         if (threads_active)
 126                 pthread_mutex_unlock(mutex);
 127 }
 128
 129 /*
 130  * Mutex and conditional variable can't be statically-initialized on Windows.
 131  */
 132 static void init_thread(void)
 133 {
 134         int i;
 135         init_recursive_mutex(&read_mutex);
 136         pthread_mutex_init(&counter_mutex, NULL);
 137         pthread_mutex_init(&work_mutex, NULL);
 138         if (show_stat)
 139                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 140         pthread_key_create(&key, NULL);
 141         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 142         for (i = 0; i < nr_threads; i++) {
 143                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 144                 if (thread_data[i].pack_fd == -1)
 145                         die_errno(_("unable to open %s"), curr_pack);
 146         }
 147
 148         threads_active = 1;
 149 }
 150
 151 static void cleanup_thread(void)
 152 {
 153         int i;
 154         if (!threads_active)
 155                 return;
 156         threads_active = 0;
 157         pthread_mutex_destroy(&read_mutex);
 158         pthread_mutex_destroy(&counter_mutex);
 159         pthread_mutex_destroy(&work_mutex);
 160         if (show_stat)
 161                 pthread_mutex_destroy(&deepest_delta_mutex);
 162         for (i = 0; i < nr_threads; i++)
 163                 close(thread_data[i].pack_fd);
 164         pthread_key_delete(key);
 165         free(thread_data);
 166 }
 167
 168 #else
 169
 170 #define read_lock()
 171 #define read_unlock()
 172
 173 #define counter_lock()
 174 #define counter_unlock()
 175
 176 #define work_lock()
 177 #define work_unlock()
 178
 179 #define deepest_delta_lock()
 180 #define deepest_delta_unlock()
 181
 182 #endif
 183
 184
 185 static int mark_link(struct object *obj, int type, void *data)
 186 {
 187         if (!obj)
 188                 return -1;
 189
 190         if (type != OBJ_ANY && obj->type != type)
 191                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 192
 193         obj->flags |= FLAG_LINK;
 194         return 0;
 195 }
 196
 197 /* The content of each linked object must have been checked
 198    or it must be already present in the object database */
 199 static unsigned check_object(struct object *obj)
 200 {
 201         if (!obj)
 202                 return 0;
 203
 204         if (!(obj->flags & FLAG_LINK))
 205                 return 0;
 206
 207         if (!(obj->flags & FLAG_CHECKED)) {
 208                 unsigned long size;
 209                 int type = sha1_object_info(obj->sha1, &size);
 210                 if (type != obj->type || type <= 0)
 211                         die(_("object of unexpected type"));
 212                 obj->flags |= FLAG_CHECKED;
 213                 return 1;
 214         }
 215
 216         return 0;
 217 }
 218
 219 static unsigned check_objects(void)
 220 {
 221         unsigned i, max, foreign_nr = 0;
 222
 223         max = get_max_object_index();
 224         for (i = 0; i < max; i++)
 225                 foreign_nr += check_object(get_indexed_object(i));
 226         return foreign_nr;
 227 }
 228
 229
 230 /* Discard current buffer used content. */
 231 static void flush(void)
 232 {
 233         if (input_offset) {
 234                 if (output_fd >= 0)
 235                         write_or_die(output_fd, input_buffer, input_offset);
 236                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 237                 memmove(input_buffer, input_buffer + input_offset, input_len);
 238                 input_offset = 0;
 239         }
 240 }
 241
 242 /*
 243  * Make sure at least "min" bytes are available in the buffer, and
 244  * return the pointer to the buffer.
 245  */
 246 static void *fill(int min)
 247 {
 248         if (min <= input_len)
 249                 return input_buffer + input_offset;
 250         if (min > sizeof(input_buffer))
 251                 die(Q_("cannot fill %d byte",
 252                        "cannot fill %d bytes",
 253                        min),
 254                     min);
 255         flush();
 256         do {
 257                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 258                                 sizeof(input_buffer) - input_len);
 259                 if (ret <= 0) {
 260                         if (!ret)
 261                                 die(_("early EOF"));
 262                         die_errno(_("read error on input"));
 263                 }
 264                 input_len += ret;
 265                 if (from_stdin)
 266                         display_throughput(progress, consumed_bytes + input_len);
 267         } while (input_len < min);
 268         return input_buffer;
 269 }
 270
 271 static void use(int bytes)
 272 {
 273         if (bytes > input_len)
 274                 die(_("used more bytes than were available"));
 275         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 276         input_len -= bytes;
 277         input_offset += bytes;
 278
 279         /* make sure off_t is sufficiently large not to wrap */
 280         if (signed_add_overflows(consumed_bytes, bytes))
 281                 die(_("pack too large for current definition of off_t"));
 282         consumed_bytes += bytes;
 283 }
 284
 285 static const char *open_pack_file(const char *pack_name)
 286 {
 287         if (from_stdin) {
 288                 input_fd = 0;
 289                 if (!pack_name) {
 290                         static char tmp_file[PATH_MAX];
 291                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 292                                                 "pack/tmp_pack_XXXXXX");
 293                         pack_name = xstrdup(tmp_file);
 294                 } else
 295                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 296                 if (output_fd < 0)
 297                         die_errno(_("unable to create '%s'"), pack_name);
 298                 nothread_data.pack_fd = output_fd;
 299         } else {
 300                 input_fd = open(pack_name, O_RDONLY);
 301                 if (input_fd < 0)
 302                         die_errno(_("cannot open packfile '%s'"), pack_name);
 303                 output_fd = -1;
 304                 nothread_data.pack_fd = input_fd;
 305         }
 306         git_SHA1_Init(&input_ctx);
 307         return pack_name;
 308 }
 309
 310 static void parse_pack_header(void)
 311 {
 312         struct pack_header *hdr = fill(sizeof(struct pack_header));
 313
 314         /* Header consistency check */
 315         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 316                 die(_("pack signature mismatch"));
 317         if (!pack_version_ok(hdr->hdr_version))
 318                 die(_("pack version %"PRIu32" unsupported"),
 319                         ntohl(hdr->hdr_version));
 320
 321         nr_objects = ntohl(hdr->hdr_entries);
 322         use(sizeof(struct pack_header));
 323 }
 324
 325 static NORETURN void bad_object(unsigned long offset, const char *format,
 326                        ...) __attribute__((format (printf, 2, 3)));
 327
 328 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 329 {
 330         va_list params;
 331         char buf[1024];
 332
 333         va_start(params, format);
 334         vsnprintf(buf, sizeof(buf), format, params);
 335         va_end(params);
 336         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 337 }
 338
 339 static inline struct thread_local *get_thread_data(void)
 340 {
 341 #ifndef NO_PTHREADS
 342         if (threads_active)
 343                 return pthread_getspecific(key);
 344         assert(!threads_active &&
 345                "This should only be reached when all threads are gone");
 346 #endif
 347         return &nothread_data;
 348 }
 349
 350 #ifndef NO_PTHREADS
 351 static void set_thread_data(struct thread_local *data)
 352 {
 353         if (threads_active)
 354                 pthread_setspecific(key, data);
 355 }
 356 #endif
 357
 358 static struct base_data *alloc_base_data(void)
 359 {
 360         struct base_data *base = xmalloc(sizeof(struct base_data));
 361         memset(base, 0, sizeof(*base));
 362         base->ref_last = -1;
 363         base->ofs_last = -1;
 364         return base;
 365 }
 366
 367 static void free_base_data(struct base_data *c)
 368 {
 369         if (c->data) {
 370                 free(c->data);
 371                 c->data = NULL;
 372                 get_thread_data()->base_cache_used -= c->size;
 373         }
 374 }
 375
 376 static void prune_base_data(struct base_data *retain)
 377 {
 378         struct base_data *b;
 379         struct thread_local *data = get_thread_data();
 380         for (b = data->base_cache;
 381              data->base_cache_used > delta_base_cache_limit && b;
 382              b = b->child) {
 383                 if (b->data && b != retain)
 384                         free_base_data(b);
 385         }
 386 }
 387
 388 static void link_base_data(struct base_data *base, struct base_data *c)
 389 {
 390         if (base)
 391                 base->child = c;
 392         else
 393                 get_thread_data()->base_cache = c;
 394
 395         c->base = base;
 396         c->child = NULL;
 397         if (c->data)
 398                 get_thread_data()->base_cache_used += c->size;
 399         prune_base_data(c);
 400 }
 401
 402 static void unlink_base_data(struct base_data *c)
 403 {
 404         struct base_data *base = c->base;
 405         if (base)
 406                 base->child = NULL;
 407         else
 408                 get_thread_data()->base_cache = NULL;
 409         free_base_data(c);
 410 }
 411
 412 static int is_delta_type(enum object_type type)
 413 {
 414         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 415 }
 416
 417 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 418                                enum object_type type, unsigned char *sha1)
 419 {
 420         static char fixed_buf[8192];
 421         int status;
 422         git_zstream stream;
 423         void *buf;
 424         git_SHA_CTX c;
 425         char hdr[32];
 426         int hdrlen;
 427
 428         if (!is_delta_type(type)) {
 429                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 430                 git_SHA1_Init(&c);
 431                 git_SHA1_Update(&c, hdr, hdrlen);
 432         } else
 433                 sha1 = NULL;
 434         if (type == OBJ_BLOB && size > big_file_threshold)
 435                 buf = fixed_buf;
 436         else
 437                 buf = xmalloc(size);
 438
 439         memset(&stream, 0, sizeof(stream));
 440         git_inflate_init(&stream);
 441         stream.next_out = buf;
 442         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 443
 444         do {
 445                 unsigned char *last_out = stream.next_out;
 446                 stream.next_in = fill(1);
 447                 stream.avail_in = input_len;
 448                 status = git_inflate(&stream, 0);
 449                 use(input_len - stream.avail_in);
 450                 if (sha1)
 451                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 452                 if (buf == fixed_buf) {
 453                         stream.next_out = buf;
 454                         stream.avail_out = sizeof(fixed_buf);
 455                 }
 456         } while (status == Z_OK);
 457         if (stream.total_out != size || status != Z_STREAM_END)
 458                 bad_object(offset, _("inflate returned %d"), status);
 459         git_inflate_end(&stream);
 460         if (sha1)
 461                 git_SHA1_Final(sha1, &c);
 462         return buf == fixed_buf ? NULL : buf;
 463 }
 464
 465 static void *unpack_raw_entry(struct object_entry *obj,
 466                               union delta_base *delta_base,
 467                               unsigned char *sha1)
 468 {
 469         unsigned char *p;
 470         unsigned long size, c;
 471         off_t base_offset;
 472         unsigned shift;
 473         void *data;
 474
 475         obj->idx.offset = consumed_bytes;
 476         input_crc32 = crc32(0, NULL, 0);
 477
 478         p = fill(1);
 479         c = *p;
 480         use(1);
 481         obj->type = (c >> 4) & 7;
 482         size = (c & 15);
 483         shift = 4;
 484         while (c & 0x80) {
 485                 p = fill(1);
 486                 c = *p;
 487                 use(1);
 488                 size += (c & 0x7f) << shift;
 489                 shift += 7;
 490         }
 491         obj->size = size;
 492
 493         switch (obj->type) {
 494         case OBJ_REF_DELTA:
 495                 hashcpy(delta_base->sha1, fill(20));
 496                 use(20);
 497                 break;
 498         case OBJ_OFS_DELTA:
 499                 memset(delta_base, 0, sizeof(*delta_base));
 500                 p = fill(1);
 501                 c = *p;
 502                 use(1);
 503                 base_offset = c & 127;
 504                 while (c & 128) {
 505                         base_offset += 1;
 506                         if (!base_offset || MSB(base_offset, 7))
 507                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 508                         p = fill(1);
 509                         c = *p;
 510                         use(1);
 511                         base_offset = (base_offset << 7) + (c & 127);
 512                 }
 513                 delta_base->offset = obj->idx.offset - base_offset;
 514                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 515                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 516                 break;
 517         case OBJ_COMMIT:
 518         case OBJ_TREE:
 519         case OBJ_BLOB:
 520         case OBJ_TAG:
 521                 break;
 522         default:
 523                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 524         }
 525         obj->hdr_size = consumed_bytes - obj->idx.offset;
 526
 527         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 528         obj->idx.crc32 = input_crc32;
 529         return data;
 530 }
 531
 532 static void *unpack_data(struct object_entry *obj,
 533                          int (*consume)(const unsigned char *, unsigned long, void *),
 534                          void *cb_data)
 535 {
 536         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 537         unsigned long len = obj[1].idx.offset - from;
 538         unsigned char *data, *inbuf;
 539         git_zstream stream;
 540         int status;
 541
 542         data = xmalloc(consume ? 64*1024 : obj->size);
 543         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 544
 545         memset(&stream, 0, sizeof(stream));
 546         git_inflate_init(&stream);
 547         stream.next_out = data;
 548         stream.avail_out = consume ? 64*1024 : obj->size;
 549
 550         do {
 551                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 552                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 553                 if (n < 0)
 554                         die_errno(_("cannot pread pack file"));
 555                 if (!n)
 556                         die(Q_("premature end of pack file, %lu byte missing",
 557                                "premature end of pack file, %lu bytes missing",
 558                                len),
 559                             len);
 560                 from += n;
 561                 len -= n;
 562                 stream.next_in = inbuf;
 563                 stream.avail_in = n;
 564                 if (!consume)
 565                         status = git_inflate(&stream, 0);
 566                 else {
 567                         do {
 568                                 status = git_inflate(&stream, 0);
 569                                 if (consume(data, stream.next_out - data, cb_data)) {
 570                                         free(inbuf);
 571                                         free(data);
 572                                         return NULL;
 573                                 }
 574                                 stream.next_out = data;
 575                                 stream.avail_out = 64*1024;
 576                         } while (status == Z_OK && stream.avail_in);
 577                 }
 578         } while (len && status == Z_OK && !stream.avail_in);
 579
 580         /* This has been inflated OK when first encountered, so... */
 581         if (status != Z_STREAM_END || stream.total_out != obj->size)
 582                 die(_("serious inflate inconsistency"));
 583
 584         git_inflate_end(&stream);
 585         free(inbuf);
 586         if (consume) {
 587                 free(data);
 588                 data = NULL;
 589         }
 590         return data;
 591 }
 592
 593 static void *get_data_from_pack(struct object_entry *obj)
 594 {
 595         return unpack_data(obj, NULL, NULL);
 596 }
 597
 598 static int compare_delta_bases(const union delta_base *base1,
 599                                const union delta_base *base2,
 600                                enum object_type type1,
 601                                enum object_type type2)
 602 {
 603         int cmp = type1 - type2;
 604         if (cmp)
 605                 return cmp;
 606         return memcmp(base1, base2, UNION_BASE_SZ);
 607 }
 608
 609 static int find_delta(const union delta_base *base, enum object_type type)
 610 {
 611         int first = 0, last = nr_deltas;
 612
 613         while (first < last) {
 614                 int next = (first + last) / 2;
 615                 struct delta_entry *delta = &deltas[next];
 616                 int cmp;
 617
 618                 cmp = compare_delta_bases(base, &delta->base,
 619                                           type, objects[delta->obj_no].type);
 620                 if (!cmp)
 621                         return next;
 622                 if (cmp < 0) {
 623                         last = next;
 624                         continue;
 625                 }
 626                 first = next+1;
 627         }
 628         return -first-1;
 629 }
 630
 631 static void find_delta_children(const union delta_base *base,
 632                                 int *first_index, int *last_index,
 633                                 enum object_type type)
 634 {
 635         int first = find_delta(base, type);
 636         int last = first;
 637         int end = nr_deltas - 1;
 638
 639         if (first < 0) {
 640                 *first_index = 0;
 641                 *last_index = -1;
 642                 return;
 643         }
 644         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 645                 --first;
 646         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 647                 ++last;
 648         *first_index = first;
 649         *last_index = last;
 650 }
 651
 652 struct compare_data {
 653         struct object_entry *entry;
 654         struct git_istream *st;
 655         unsigned char *buf;
 656         unsigned long buf_size;
 657 };
 658
 659 static int compare_objects(const unsigned char *buf, unsigned long size,
 660                            void *cb_data)
 661 {
 662         struct compare_data *data = cb_data;
 663
 664         if (data->buf_size < size) {
 665                 free(data->buf);
 666                 data->buf = xmalloc(size);
 667                 data->buf_size = size;
 668         }
 669
 670         while (size) {
 671                 ssize_t len = read_istream(data->st, data->buf, size);
 672                 if (len == 0)
 673                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 674                             sha1_to_hex(data->entry->idx.sha1));
 675                 if (len < 0)
 676                         die(_("unable to read %s"),
 677                             sha1_to_hex(data->entry->idx.sha1));
 678                 if (memcmp(buf, data->buf, len))
 679                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 680                             sha1_to_hex(data->entry->idx.sha1));
 681                 size -= len;
 682                 buf += len;
 683         }
 684         return 0;
 685 }
 686
 687 static int check_collison(struct object_entry *entry)
 688 {
 689         struct compare_data data;
 690         enum object_type type;
 691         unsigned long size;
 692
 693         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 694                 return -1;
 695
 696         memset(&data, 0, sizeof(data));
 697         data.entry = entry;
 698         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 699         if (!data.st)
 700                 return -1;
 701         if (size != entry->size || type != entry->type)
 702                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 703                     sha1_to_hex(entry->idx.sha1));
 704         unpack_data(entry, compare_objects, &data);
 705         close_istream(data.st);
 706         free(data.buf);
 707         return 0;
 708 }
 709
 710 static void sha1_object(const void *data, struct object_entry *obj_entry,
 711                         unsigned long size, enum object_type type,
 712                         const unsigned char *sha1)
 713 {
 714         void *new_data = NULL;
 715         int collision_test_needed;
 716
 717         assert(data || obj_entry);
 718
 719         read_lock();
 720         collision_test_needed = has_sha1_file(sha1);
 721         read_unlock();
 722
 723         if (collision_test_needed && !data) {
 724                 read_lock();
 725                 if (!check_collison(obj_entry))
 726                         collision_test_needed = 0;
 727                 read_unlock();
 728         }
 729         if (collision_test_needed) {
 730                 void *has_data;
 731                 enum object_type has_type;
 732                 unsigned long has_size;
 733                 read_lock();
 734                 has_type = sha1_object_info(sha1, &has_size);
 735                 if (has_type != type || has_size != size)
 736                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 737                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 738                 read_unlock();
 739                 if (!data)
 740                         data = new_data = get_data_from_pack(obj_entry);
 741                 if (!has_data)
 742                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 743                 if (size != has_size || type != has_type ||
 744                     memcmp(data, has_data, size) != 0)
 745                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 746                 free(has_data);
 747         }
 748
 749         if (strict) {
 750                 read_lock();
 751                 if (type == OBJ_BLOB) {
 752                         struct blob *blob = lookup_blob(sha1);
 753                         if (blob)
 754                                 blob->object.flags |= FLAG_CHECKED;
 755                         else
 756                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 757                 } else {
 758                         struct object *obj;
 759                         int eaten;
 760                         void *buf = (void *) data;
 761
 762                         assert(data && "data can only be NULL for large _blobs_");
 763
 764                         /*
 765                          * we do not need to free the memory here, as the
 766                          * buf is deleted by the caller.
 767                          */
 768                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 769                         if (!obj)
 770                                 die(_("invalid %s"), typename(type));
 771                         if (do_fsck_object &&
 772                             fsck_object(obj, 1, fsck_error_function))
 773                                 die(_("Error in object"));
 774                         if (fsck_walk(obj, mark_link, NULL))
 775                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 776
 777                         if (obj->type == OBJ_TREE) {
 778                                 struct tree *item = (struct tree *) obj;
 779                                 item->buffer = NULL;
 780                                 obj->parsed = 0;
 781                         }
 782                         if (obj->type == OBJ_COMMIT) {
 783                                 struct commit *commit = (struct commit *) obj;
 784                                 commit->buffer = NULL;
 785                         }
 786                         obj->flags |= FLAG_CHECKED;
 787                 }
 788                 read_unlock();
 789         }
 790
 791         free(new_data);
 792 }
 793
 794 /*
 795  * This function is part of find_unresolved_deltas(). There are two
 796  * walkers going in the opposite ways.
 797  *
 798  * The first one in find_unresolved_deltas() traverses down from
 799  * parent node to children, deflating nodes along the way. However,
 800  * memory for deflated nodes is limited by delta_base_cache_limit, so
 801  * at some point parent node's deflated content may be freed.
 802  *
 803  * The second walker is this function, which goes from current node up
 804  * to top parent if necessary to deflate the node. In normal
 805  * situation, its parent node would be already deflated, so it just
 806  * needs to apply delta.
 807  *
 808  * In the worst case scenario, parent node is no longer deflated because
 809  * we're running out of delta_base_cache_limit; we need to re-deflate
 810  * parents, possibly up to the top base.
 811  *
 812  * All deflated objects here are subject to be freed if we exceed
 813  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 814  * just need to make sure the last node is not freed.
 815  */
 816 static void *get_base_data(struct base_data *c)
 817 {
 818         if (!c->data) {
 819                 struct object_entry *obj = c->obj;
 820                 struct base_data **delta = NULL;
 821                 int delta_nr = 0, delta_alloc = 0;
 822
 823                 while (is_delta_type(c->obj->type) && !c->data) {
 824                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 825                         delta[delta_nr++] = c;
 826                         c = c->base;
 827                 }
 828                 if (!delta_nr) {
 829                         c->data = get_data_from_pack(obj);
 830                         c->size = obj->size;
 831                         get_thread_data()->base_cache_used += c->size;
 832                         prune_base_data(c);
 833                 }
 834                 for (; delta_nr > 0; delta_nr--) {
 835                         void *base, *raw;
 836                         c = delta[delta_nr - 1];
 837                         obj = c->obj;
 838                         base = get_base_data(c->base);
 839                         raw = get_data_from_pack(obj);
 840                         c->data = patch_delta(
 841                                 base, c->base->size,
 842                                 raw, obj->size,
 843                                 &c->size);
 844                         free(raw);
 845                         if (!c->data)
 846                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 847                         get_thread_data()->base_cache_used += c->size;
 848                         prune_base_data(c);
 849                 }
 850                 free(delta);
 851         }
 852         return c->data;
 853 }
 854
 855 static void resolve_delta(struct object_entry *delta_obj,
 856                           struct base_data *base, struct base_data *result)
 857 {
 858         void *base_data, *delta_data;
 859
 860         delta_obj->real_type = base->obj->real_type;
 861         if (show_stat) {
 862                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 863                 deepest_delta_lock();
 864                 if (deepest_delta < delta_obj->delta_depth)
 865                         deepest_delta = delta_obj->delta_depth;
 866                 deepest_delta_unlock();
 867         }
 868         delta_obj->base_object_no = base->obj - objects;
 869         delta_data = get_data_from_pack(delta_obj);
 870         base_data = get_base_data(base);
 871         result->obj = delta_obj;
 872         result->data = patch_delta(base_data, base->size,
 873                                    delta_data, delta_obj->size, &result->size);
 874         free(delta_data);
 875         if (!result->data)
 876                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 877         hash_sha1_file(result->data, result->size,
 878                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 879         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 880                     delta_obj->idx.sha1);
 881         counter_lock();
 882         nr_resolved_deltas++;
 883         counter_unlock();
 884 }
 885
 886 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 887                                                   struct base_data *prev_base)
 888 {
 889         if (base->ref_last == -1 && base->ofs_last == -1) {
 890                 union delta_base base_spec;
 891
 892                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 893                 find_delta_children(&base_spec,
 894                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 895
 896                 memset(&base_spec, 0, sizeof(base_spec));
 897                 base_spec.offset = base->obj->idx.offset;
 898                 find_delta_children(&base_spec,
 899                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 900
 901                 if (base->ref_last == -1 && base->ofs_last == -1) {
 902                         free(base->data);
 903                         return NULL;
 904                 }
 905
 906                 link_base_data(prev_base, base);
 907         }
 908
 909         if (base->ref_first <= base->ref_last) {
 910                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 911                 struct base_data *result = alloc_base_data();
 912
 913                 assert(child->real_type == OBJ_REF_DELTA);
 914                 resolve_delta(child, base, result);
 915                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 916                         free_base_data(base);
 917
 918                 base->ref_first++;
 919                 return result;
 920         }
 921
 922         if (base->ofs_first <= base->ofs_last) {
 923                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 924                 struct base_data *result = alloc_base_data();
 925
 926                 assert(child->real_type == OBJ_OFS_DELTA);
 927                 resolve_delta(child, base, result);
 928                 if (base->ofs_first == base->ofs_last)
 929                         free_base_data(base);
 930
 931                 base->ofs_first++;
 932                 return result;
 933         }
 934
 935         unlink_base_data(base);
 936         return NULL;
 937 }
 938
 939 static void find_unresolved_deltas(struct base_data *base)
 940 {
 941         struct base_data *new_base, *prev_base = NULL;
 942         for (;;) {
 943                 new_base = find_unresolved_deltas_1(base, prev_base);
 944
 945                 if (new_base) {
 946                         prev_base = base;
 947                         base = new_base;
 948                 } else {
 949                         free(base);
 950                         base = prev_base;
 951                         if (!base)
 952                                 return;
 953                         prev_base = base->base;
 954                 }
 955         }
 956 }
 957
 958 static int compare_delta_entry(const void *a, const void *b)
 959 {
 960         const struct delta_entry *delta_a = a;
 961         const struct delta_entry *delta_b = b;
 962
 963         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 964         return compare_delta_bases(&delta_a->base, &delta_b->base,
 965                                    objects[delta_a->obj_no].type,
 966                                    objects[delta_b->obj_no].type);
 967 }
 968
 969 static void resolve_base(struct object_entry *obj)
 970 {
 971         struct base_data *base_obj = alloc_base_data();
 972         base_obj->obj = obj;
 973         base_obj->data = NULL;
 974         find_unresolved_deltas(base_obj);
 975 }
 976
 977 #ifndef NO_PTHREADS
 978 static void *threaded_second_pass(void *data)
 979 {
 980         set_thread_data(data);
 981         for (;;) {
 982                 int i;
 983                 counter_lock();
 984                 display_progress(progress, nr_resolved_deltas);
 985                 counter_unlock();
 986                 work_lock();
 987                 while (nr_dispatched < nr_objects &&
 988                        is_delta_type(objects[nr_dispatched].type))
 989                         nr_dispatched++;
 990                 if (nr_dispatched >= nr_objects) {
 991                         work_unlock();
 992                         break;
 993                 }
 994                 i = nr_dispatched++;
 995                 work_unlock();
 996
 997                 resolve_base(&objects[i]);
 998         }
 999         return NULL;
1000 }
1001 #endif
1002
1003 /*
1004  * First pass:
1005  * - find locations of all objects;
1006  * - calculate SHA1 of all non-delta objects;
1007  * - remember base (SHA1 or offset) for all deltas.
1008  */
1009 static void parse_pack_objects(unsigned char *sha1)
1010 {
1011         int i, nr_delays = 0;
1012         struct delta_entry *delta = deltas;
1013         struct stat st;
1014
1015         if (verbose)
1016                 progress = start_progress(
1017                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1018                                 nr_objects);
1019         for (i = 0; i < nr_objects; i++) {
1020                 struct object_entry *obj = &objects[i];
1021                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1022                 obj->real_type = obj->type;
1023                 if (is_delta_type(obj->type)) {
1024                         nr_deltas++;
1025                         delta->obj_no = i;
1026                         delta++;
1027                 } else if (!data) {
1028                         /* large blobs, check later */
1029                         obj->real_type = OBJ_BAD;
1030                         nr_delays++;
1031                 } else
1032                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1033                 free(data);
1034                 display_progress(progress, i+1);
1035         }
1036         objects[i].idx.offset = consumed_bytes;
1037         stop_progress(&progress);
1038
1039         /* Check pack integrity */
1040         flush();
1041         git_SHA1_Final(sha1, &input_ctx);
1042         if (hashcmp(fill(20), sha1))
1043                 die(_("pack is corrupted (SHA1 mismatch)"));
1044         use(20);
1045
1046         /* If input_fd is a file, we should have reached its end now. */
1047         if (fstat(input_fd, &st))
1048                 die_errno(_("cannot fstat packfile"));
1049         if (S_ISREG(st.st_mode) &&
1050                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1051                 die(_("pack has junk at the end"));
1052
1053         for (i = 0; i < nr_objects; i++) {
1054                 struct object_entry *obj = &objects[i];
1055                 if (obj->real_type != OBJ_BAD)
1056                         continue;
1057                 obj->real_type = obj->type;
1058                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1059                 nr_delays--;
1060         }
1061         if (nr_delays)
1062                 die(_("confusion beyond insanity in parse_pack_objects()"));
1063 }
1064
1065 /*
1066  * Second pass:
1067  * - for all non-delta objects, look if it is used as a base for
1068  *   deltas;
1069  * - if used as a base, uncompress the object and apply all deltas,
1070  *   recursively checking if the resulting object is used as a base
1071  *   for some more deltas.
1072  */
1073 static void resolve_deltas(void)
1074 {
1075         int i;
1076
1077         if (!nr_deltas)
1078                 return;
1079
1080         /* Sort deltas by base SHA1/offset for fast searching */
1081         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1082               compare_delta_entry);
1083
1084         if (verbose)
1085                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1086
1087 #ifndef NO_PTHREADS
1088         nr_dispatched = 0;
1089         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1090                 init_thread();
1091                 for (i = 0; i < nr_threads; i++) {
1092                         int ret = pthread_create(&thread_data[i].thread, NULL,
1093                                                  threaded_second_pass, thread_data + i);
1094                         if (ret)
1095                                 die(_("unable to create thread: %s"),
1096                                     strerror(ret));
1097                 }
1098                 for (i = 0; i < nr_threads; i++)
1099                         pthread_join(thread_data[i].thread, NULL);
1100                 cleanup_thread();
1101                 return;
1102         }
1103 #endif
1104
1105         for (i = 0; i < nr_objects; i++) {
1106                 struct object_entry *obj = &objects[i];
1107
1108                 if (is_delta_type(obj->type))
1109                         continue;
1110                 resolve_base(obj);
1111                 display_progress(progress, nr_resolved_deltas);
1112         }
1113 }
1114
1115 /*
1116  * Third pass:
1117  * - append objects to convert thin pack to full pack if required
1118  * - write the final 20-byte SHA-1
1119  */
1120 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1121 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1122 {
1123         if (nr_deltas == nr_resolved_deltas) {
1124                 stop_progress(&progress);
1125                 /* Flush remaining pack final 20-byte SHA1. */
1126                 flush();
1127                 return;
1128         }
1129
1130         if (fix_thin_pack) {
1131                 struct sha1file *f;
1132                 unsigned char read_sha1[20], tail_sha1[20];
1133                 struct strbuf msg = STRBUF_INIT;
1134                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1135                 int nr_objects_initial = nr_objects;
1136                 if (nr_unresolved <= 0)
1137                         die(_("confusion beyond insanity"));
1138                 objects = xrealloc(objects,
1139                                    (nr_objects + nr_unresolved + 1)
1140                                    * sizeof(*objects));
1141                 memset(objects + nr_objects + 1, 0,
1142                        nr_unresolved * sizeof(*objects));
1143                 f = sha1fd(output_fd, curr_pack);
1144                 fix_unresolved_deltas(f, nr_unresolved);
1145                 strbuf_addf(&msg, _("completed with %d local objects"),
1146                             nr_objects - nr_objects_initial);
1147                 stop_progress_msg(&progress, msg.buf);
1148                 strbuf_release(&msg);
1149                 sha1close(f, tail_sha1, 0);
1150                 hashcpy(read_sha1, pack_sha1);
1151                 fixup_pack_header_footer(output_fd, pack_sha1,
1152                                          curr_pack, nr_objects,
1153                                          read_sha1, consumed_bytes-20);
1154                 if (hashcmp(read_sha1, tail_sha1) != 0)
1155                         die(_("Unexpected tail checksum for %s "
1156                               "(disk corruption?)"), curr_pack);
1157         }
1158         if (nr_deltas != nr_resolved_deltas)
1159                 die(Q_("pack has %d unresolved delta",
1160                        "pack has %d unresolved deltas",
1161                        nr_deltas - nr_resolved_deltas),
1162                     nr_deltas - nr_resolved_deltas);
1163 }
1164
1165 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1166 {
1167         git_zstream stream;
1168         int status;
1169         unsigned char outbuf[4096];
1170
1171         memset(&stream, 0, sizeof(stream));
1172         git_deflate_init(&stream, zlib_compression_level);
1173         stream.next_in = in;
1174         stream.avail_in = size;
1175
1176         do {
1177                 stream.next_out = outbuf;
1178                 stream.avail_out = sizeof(outbuf);
1179                 status = git_deflate(&stream, Z_FINISH);
1180                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1181         } while (status == Z_OK);
1182
1183         if (status != Z_STREAM_END)
1184                 die(_("unable to deflate appended object (%d)"), status);
1185         size = stream.total_out;
1186         git_deflate_end(&stream);
1187         return size;
1188 }
1189
1190 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1191                                const unsigned char *sha1, void *buf,
1192                                unsigned long size, enum object_type type)
1193 {
1194         struct object_entry *obj = &objects[nr_objects++];
1195         unsigned char header[10];
1196         unsigned long s = size;
1197         int n = 0;
1198         unsigned char c = (type << 4) | (s & 15);
1199         s >>= 4;
1200         while (s) {
1201                 header[n++] = c | 0x80;
1202                 c = s & 0x7f;
1203                 s >>= 7;
1204         }
1205         header[n++] = c;
1206         crc32_begin(f);
1207         sha1write(f, header, n);
1208         obj[0].size = size;
1209         obj[0].hdr_size = n;
1210         obj[0].type = type;
1211         obj[0].real_type = type;
1212         obj[1].idx.offset = obj[0].idx.offset + n;
1213         obj[1].idx.offset += write_compressed(f, buf, size);
1214         obj[0].idx.crc32 = crc32_end(f);
1215         sha1flush(f);
1216         hashcpy(obj->idx.sha1, sha1);
1217         return obj;
1218 }
1219
1220 static int delta_pos_compare(const void *_a, const void *_b)
1221 {
1222         struct delta_entry *a = *(struct delta_entry **)_a;
1223         struct delta_entry *b = *(struct delta_entry **)_b;
1224         return a->obj_no - b->obj_no;
1225 }
1226
1227 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1228 {
1229         struct delta_entry **sorted_by_pos;
1230         int i, n = 0;
1231
1232         /*
1233          * Since many unresolved deltas may well be themselves base objects
1234          * for more unresolved deltas, we really want to include the
1235          * smallest number of base objects that would cover as much delta
1236          * as possible by picking the
1237          * trunc deltas first, allowing for other deltas to resolve without
1238          * additional base objects.  Since most base objects are to be found
1239          * before deltas depending on them, a good heuristic is to start
1240          * resolving deltas in the same order as their position in the pack.
1241          */
1242         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1243         for (i = 0; i < nr_deltas; i++) {
1244                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1245                         continue;
1246                 sorted_by_pos[n++] = &deltas[i];
1247         }
1248         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1249
1250         for (i = 0; i < n; i++) {
1251                 struct delta_entry *d = sorted_by_pos[i];
1252                 enum object_type type;
1253                 struct base_data *base_obj = alloc_base_data();
1254
1255                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1256                         continue;
1257                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1258                 if (!base_obj->data)
1259                         continue;
1260
1261                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1262                                 base_obj->size, typename(type)))
1263                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1264                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1265                                         base_obj->data, base_obj->size, type);
1266                 find_unresolved_deltas(base_obj);
1267                 display_progress(progress, nr_resolved_deltas);
1268         }
1269         free(sorted_by_pos);
1270 }
1271
1272 static void final(const char *final_pack_name, const char *curr_pack_name,
1273                   const char *final_index_name, const char *curr_index_name,
1274                   const char *keep_name, const char *keep_msg,
1275                   unsigned char *sha1)
1276 {
1277         const char *report = "pack";
1278         char name[PATH_MAX];
1279         int err;
1280
1281         if (!from_stdin) {
1282                 close(input_fd);
1283         } else {
1284                 fsync_or_die(output_fd, curr_pack_name);
1285                 err = close(output_fd);
1286                 if (err)
1287                         die_errno(_("error while closing pack file"));
1288         }
1289
1290         if (keep_msg) {
1291                 int keep_fd, keep_msg_len = strlen(keep_msg);
1292
1293                 if (!keep_name)
1294                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1295                 else
1296                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1297
1298                 if (keep_fd < 0) {
1299                         if (errno != EEXIST)
1300                                 die_errno(_("cannot write keep file '%s'"),
1301                                           keep_name ? keep_name : name);
1302                 } else {
1303                         if (keep_msg_len > 0) {
1304                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1305                                 write_or_die(keep_fd, "\n", 1);
1306                         }
1307                         if (close(keep_fd) != 0)
1308                                 die_errno(_("cannot close written keep file '%s'"),
1309                                           keep_name ? keep_name : name);
1310                         report = "keep";
1311                 }
1312         }
1313
1314         if (final_pack_name != curr_pack_name) {
1315                 if (!final_pack_name) {
1316                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1317                                  get_object_directory(), sha1_to_hex(sha1));
1318                         final_pack_name = name;
1319                 }
1320                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1321                         die(_("cannot store pack file"));
1322         } else if (from_stdin)
1323                 chmod(final_pack_name, 0444);
1324
1325         if (final_index_name != curr_index_name) {
1326                 if (!final_index_name) {
1327                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1328                                  get_object_directory(), sha1_to_hex(sha1));
1329                         final_index_name = name;
1330                 }
1331                 if (move_temp_to_file(curr_index_name, final_index_name))
1332                         die(_("cannot store index file"));
1333         } else
1334                 chmod(final_index_name, 0444);
1335
1336         if (!from_stdin) {
1337                 printf("%s\n", sha1_to_hex(sha1));
1338         } else {
1339                 char buf[48];
1340                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1341                                    report, sha1_to_hex(sha1));
1342                 write_or_die(1, buf, len);
1343
1344                 /*
1345                  * Let's just mimic git-unpack-objects here and write
1346                  * the last part of the input buffer to stdout.
1347                  */
1348                 while (input_len) {
1349                         err = xwrite(1, input_buffer + input_offset, input_len);
1350                         if (err <= 0)
1351                                 break;
1352                         input_len -= err;
1353                         input_offset += err;
1354                 }
1355         }
1356 }
1357
1358 static int git_index_pack_config(const char *k, const char *v, void *cb)
1359 {
1360         struct pack_idx_option *opts = cb;
1361
1362         if (!strcmp(k, "pack.indexversion")) {
1363                 opts->version = git_config_int(k, v);
1364                 if (opts->version > 2)
1365                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1366                 return 0;
1367         }
1368         if (!strcmp(k, "pack.threads")) {
1369                 nr_threads = git_config_int(k, v);
1370                 if (nr_threads < 0)
1371                         die(_("invalid number of threads specified (%d)"),
1372                             nr_threads);
1373 #ifdef NO_PTHREADS
1374                 if (nr_threads != 1)
1375                         warning(_("no threads support, ignoring %s"), k);
1376                 nr_threads = 1;
1377 #endif
1378                 return 0;
1379         }
1380         return git_default_config(k, v, cb);
1381 }
1382
1383 static int cmp_uint32(const void *a_, const void *b_)
1384 {
1385         uint32_t a = *((uint32_t *)a_);
1386         uint32_t b = *((uint32_t *)b_);
1387
1388         return (a < b) ? -1 : (a != b);
1389 }
1390
1391 static void read_v2_anomalous_offsets(struct packed_git *p,
1392                                       struct pack_idx_option *opts)
1393 {
1394         const uint32_t *idx1, *idx2;
1395         uint32_t i;
1396
1397         /* The address of the 4-byte offset table */
1398         idx1 = (((const uint32_t *)p->index_data)
1399                 + 2 /* 8-byte header */
1400                 + 256 /* fan out */
1401                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1402                 + p->num_objects /* CRC32 table */
1403                 );
1404
1405         /* The address of the 8-byte offset table */
1406         idx2 = idx1 + p->num_objects;
1407
1408         for (i = 0; i < p->num_objects; i++) {
1409                 uint32_t off = ntohl(idx1[i]);
1410                 if (!(off & 0x80000000))
1411                         continue;
1412                 off = off & 0x7fffffff;
1413                 if (idx2[off * 2])
1414                         continue;
1415                 /*
1416                  * The real offset is ntohl(idx2[off * 2]) in high 4
1417                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1418                  * octets.  But idx2[off * 2] is Zero!!!
1419                  */
1420                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1421                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1422         }
1423
1424         if (1 < opts->anomaly_nr)
1425                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1426 }
1427
1428 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1429 {
1430         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1431
1432         if (!p)
1433                 die(_("Cannot open existing pack file '%s'"), pack_name);
1434         if (open_pack_index(p))
1435                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1436
1437         /* Read the attributes from the existing idx file */
1438         opts->version = p->index_version;
1439
1440         if (opts->version == 2)
1441                 read_v2_anomalous_offsets(p, opts);
1442
1443         /*
1444          * Get rid of the idx file as we do not need it anymore.
1445          * NEEDSWORK: extract this bit from free_pack_by_name() in
1446          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1447          * know we haven't installed this pack (hence we never have
1448          * read anything from it).
1449          */
1450         close_pack_index(p);
1451         free(p);
1452 }
1453
1454 static void show_pack_info(int stat_only)
1455 {
1456         int i, baseobjects = nr_objects - nr_deltas;
1457         unsigned long *chain_histogram = NULL;
1458
1459         if (deepest_delta)
1460                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1461
1462         for (i = 0; i < nr_objects; i++) {
1463                 struct object_entry *obj = &objects[i];
1464
1465                 if (is_delta_type(obj->type))
1466                         chain_histogram[obj->delta_depth - 1]++;
1467                 if (stat_only)
1468                         continue;
1469                 printf("%s %-6s %lu %lu %"PRIuMAX,
1470                        sha1_to_hex(obj->idx.sha1),
1471                        typename(obj->real_type), obj->size,
1472                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1473                        (uintmax_t)obj->idx.offset);
1474                 if (is_delta_type(obj->type)) {
1475                         struct object_entry *bobj = &objects[obj->base_object_no];
1476                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1477                 }
1478                 putchar('\n');
1479         }
1480
1481         if (baseobjects)
1482                 printf_ln(Q_("non delta: %d object",
1483                              "non delta: %d objects",
1484                              baseobjects),
1485                           baseobjects);
1486         for (i = 0; i < deepest_delta; i++) {
1487                 if (!chain_histogram[i])
1488                         continue;
1489                 printf_ln(Q_("chain length = %d: %lu object",
1490                              "chain length = %d: %lu objects",
1491                              chain_histogram[i]),
1492                           i + 1,
1493                           chain_histogram[i]);
1494         }
1495 }
1496
1497 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1498 {
1499         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1500         const char *curr_index;
1501         const char *index_name = NULL, *pack_name = NULL;
1502         const char *keep_name = NULL, *keep_msg = NULL;
1503         char *index_name_buf = NULL, *keep_name_buf = NULL;
1504         struct pack_idx_entry **idx_objects;
1505         struct pack_idx_option opts;
1506         unsigned char pack_sha1[20];
1507         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1508
1509         if (argc == 2 && !strcmp(argv[1], "-h"))
1510                 usage(index_pack_usage);
1511
1512         check_replace_refs = 0;
1513
1514         reset_pack_idx_option(&opts);
1515         git_config(git_index_pack_config, &opts);
1516         if (prefix && chdir(prefix))
1517                 die(_("Cannot come back to cwd"));
1518
1519         for (i = 1; i < argc; i++) {
1520                 const char *arg = argv[i];
1521
1522                 if (*arg == '-') {
1523                         if (!strcmp(arg, "--stdin")) {
1524                                 from_stdin = 1;
1525                         } else if (!strcmp(arg, "--fix-thin")) {
1526                                 fix_thin_pack = 1;
1527                         } else if (!strcmp(arg, "--strict")) {
1528                                 strict = 1;
1529                                 do_fsck_object = 1;
1530                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1531                                 strict = 1;
1532                                 check_self_contained_and_connected = 1;
1533                         } else if (!strcmp(arg, "--verify")) {
1534                                 verify = 1;
1535                         } else if (!strcmp(arg, "--verify-stat")) {
1536                                 verify = 1;
1537                                 show_stat = 1;
1538                         } else if (!strcmp(arg, "--verify-stat-only")) {
1539                                 verify = 1;
1540                                 show_stat = 1;
1541                                 stat_only = 1;
1542                         } else if (!strcmp(arg, "--keep")) {
1543                                 keep_msg = "";
1544                         } else if (starts_with(arg, "--keep=")) {
1545                                 keep_msg = arg + 7;
1546                         } else if (starts_with(arg, "--threads=")) {
1547                                 char *end;
1548                                 nr_threads = strtoul(arg+10, &end, 0);
1549                                 if (!arg[10] || *end || nr_threads < 0)
1550                                         usage(index_pack_usage);
1551 #ifdef NO_PTHREADS
1552                                 if (nr_threads != 1)
1553                                         warning(_("no threads support, "
1554                                                   "ignoring %s"), arg);
1555                                 nr_threads = 1;
1556 #endif
1557                         } else if (starts_with(arg, "--pack_header=")) {
1558                                 struct pack_header *hdr;
1559                                 char *c;
1560
1561                                 hdr = (struct pack_header *)input_buffer;
1562                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1563                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1564                                 if (*c != ',')
1565                                         die(_("bad %s"), arg);
1566                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1567                                 if (*c)
1568                                         die(_("bad %s"), arg);
1569                                 input_len = sizeof(*hdr);
1570                         } else if (!strcmp(arg, "-v")) {
1571                                 verbose = 1;
1572                         } else if (!strcmp(arg, "-o")) {
1573                                 if (index_name || (i+1) >= argc)
1574                                         usage(index_pack_usage);
1575                                 index_name = argv[++i];
1576                         } else if (starts_with(arg, "--index-version=")) {
1577                                 char *c;
1578                                 opts.version = strtoul(arg + 16, &c, 10);
1579                                 if (opts.version > 2)
1580                                         die(_("bad %s"), arg);
1581                                 if (*c == ',')
1582                                         opts.off32_limit = strtoul(c+1, &c, 0);
1583                                 if (*c || opts.off32_limit & 0x80000000)
1584                                         die(_("bad %s"), arg);
1585                         } else
1586                                 usage(index_pack_usage);
1587                         continue;
1588                 }
1589
1590                 if (pack_name)
1591                         usage(index_pack_usage);
1592                 pack_name = arg;
1593         }
1594
1595         if (!pack_name && !from_stdin)
1596                 usage(index_pack_usage);
1597         if (fix_thin_pack && !from_stdin)
1598                 die(_("--fix-thin cannot be used without --stdin"));
1599         if (!index_name && pack_name) {
1600                 int len = strlen(pack_name);
1601                 if (!has_extension(pack_name, ".pack"))
1602                         die(_("packfile name '%s' does not end with '.pack'"),
1603                             pack_name);
1604                 index_name_buf = xmalloc(len);
1605                 memcpy(index_name_buf, pack_name, len - 5);
1606                 strcpy(index_name_buf + len - 5, ".idx");
1607                 index_name = index_name_buf;
1608         }
1609         if (keep_msg && !keep_name && pack_name) {
1610                 int len = strlen(pack_name);
1611                 if (!has_extension(pack_name, ".pack"))
1612                         die(_("packfile name '%s' does not end with '.pack'"),
1613                             pack_name);
1614                 keep_name_buf = xmalloc(len);
1615                 memcpy(keep_name_buf, pack_name, len - 5);
1616                 strcpy(keep_name_buf + len - 5, ".keep");
1617                 keep_name = keep_name_buf;
1618         }
1619         if (verify) {
1620                 if (!index_name)
1621                         die(_("--verify with no packfile name given"));
1622                 read_idx_option(&opts, index_name);
1623                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1624         }
1625         if (strict)
1626                 opts.flags |= WRITE_IDX_STRICT;
1627
1628 #ifndef NO_PTHREADS
1629         if (!nr_threads) {
1630                 nr_threads = online_cpus();
1631                 /* An experiment showed that more threads does not mean faster */
1632                 if (nr_threads > 3)
1633                         nr_threads = 3;
1634         }
1635 #endif
1636
1637         curr_pack = open_pack_file(pack_name);
1638         parse_pack_header();
1639         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1640         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1641         parse_pack_objects(pack_sha1);
1642         resolve_deltas();
1643         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1644         free(deltas);
1645         if (strict)
1646                 foreign_nr = check_objects();
1647
1648         if (show_stat)
1649                 show_pack_info(stat_only);
1650
1651         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1652         for (i = 0; i < nr_objects; i++)
1653                 idx_objects[i] = &objects[i].idx;
1654         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1655         free(idx_objects);
1656
1657         if (!verify)
1658                 final(pack_name, curr_pack,
1659                       index_name, curr_index,
1660                       keep_name, keep_msg,
1661                       pack_sha1);
1662         else
1663                 close(input_fd);
1664         free(objects);
1665         free(index_name_buf);
1666         free(keep_name_buf);
1667         if (pack_name == NULL)
1668                 free((void *) curr_pack);
1669         if (index_name == NULL)
1670                 free((void *) curr_index);
1671
1672         /*
1673          * Let the caller know this pack is not self contained
1674          */
1675         if (check_self_contained_and_connected && foreign_nr)
1676                 return 1;
1677
1678         return 0;
1679 }