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

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned char hdr_size;
  22         signed char type;
  23         signed char real_type;
  24 };
  25
  26 struct object_stat {
  27         unsigned delta_depth;
  28         int base_object_no;
  29 };
  30
  31 struct base_data {
  32         struct base_data *base;
  33         struct base_data *child;
  34         struct object_entry *obj;
  35         void *data;
  36         unsigned long size;
  37         int ref_first, ref_last;
  38         int ofs_first, ofs_last;
  39 };
  40
  41 struct thread_local {
  42 #ifndef NO_PTHREADS
  43         pthread_t thread;
  44 #endif
  45         struct base_data *base_cache;
  46         size_t base_cache_used;
  47         int pack_fd;
  48 };
  49
  50 #define FLAG_LINK (1u<<20)
  51 #define FLAG_CHECKED (1u<<21)
  52
  53 struct ofs_delta_entry {
  54         off_t offset;
  55         int obj_no;
  56 };
  57
  58 struct ref_delta_entry {
  59         unsigned char sha1[20];
  60         int obj_no;
  61 };
  62
  63 static struct object_entry *objects;
  64 static struct object_stat *obj_stat;
  65 static struct ofs_delta_entry *ofs_deltas;
  66 static struct ref_delta_entry *ref_deltas;
  67 static struct thread_local nothread_data;
  68 static int nr_objects;
  69 static int nr_ofs_deltas;
  70 static int nr_ref_deltas;
  71 static int ref_deltas_alloc;
  72 static int nr_resolved_deltas;
  73 static int nr_threads;
  74
  75 static int from_stdin;
  76 static int strict;
  77 static int do_fsck_object;
  78 static int verbose;
  79 static int show_stat;
  80 static int check_self_contained_and_connected;
  81
  82 static struct progress *progress;
  83
  84 /* We always read in 4kB chunks. */
  85 static unsigned char input_buffer[4096];
  86 static unsigned int input_offset, input_len;
  87 static off_t consumed_bytes;
  88 static unsigned deepest_delta;
  89 static git_SHA_CTX input_ctx;
  90 static uint32_t input_crc32;
  91 static int input_fd, output_fd;
  92 static const char *curr_pack;
  93
  94 #ifndef NO_PTHREADS
  95
  96 static struct thread_local *thread_data;
  97 static int nr_dispatched;
  98 static int threads_active;
  99
 100 static pthread_mutex_t read_mutex;
 101 #define read_lock()             lock_mutex(&read_mutex)
 102 #define read_unlock()           unlock_mutex(&read_mutex)
 103
 104 static pthread_mutex_t counter_mutex;
 105 #define counter_lock()          lock_mutex(&counter_mutex)
 106 #define counter_unlock()        unlock_mutex(&counter_mutex)
 107
 108 static pthread_mutex_t work_mutex;
 109 #define work_lock()             lock_mutex(&work_mutex)
 110 #define work_unlock()           unlock_mutex(&work_mutex)
 111
 112 static pthread_mutex_t deepest_delta_mutex;
 113 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 114 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 115
 116 static pthread_mutex_t type_cas_mutex;
 117 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 118 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 119
 120 static pthread_key_t key;
 121
 122 static inline void lock_mutex(pthread_mutex_t *mutex)
 123 {
 124         if (threads_active)
 125                 pthread_mutex_lock(mutex);
 126 }
 127
 128 static inline void unlock_mutex(pthread_mutex_t *mutex)
 129 {
 130         if (threads_active)
 131                 pthread_mutex_unlock(mutex);
 132 }
 133
 134 /*
 135  * Mutex and conditional variable can't be statically-initialized on Windows.
 136  */
 137 static void init_thread(void)
 138 {
 139         int i;
 140         init_recursive_mutex(&read_mutex);
 141         pthread_mutex_init(&counter_mutex, NULL);
 142         pthread_mutex_init(&work_mutex, NULL);
 143         pthread_mutex_init(&type_cas_mutex, NULL);
 144         if (show_stat)
 145                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 146         pthread_key_create(&key, NULL);
 147         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 148         for (i = 0; i < nr_threads; i++) {
 149                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 150                 if (thread_data[i].pack_fd == -1)
 151                         die_errno(_("unable to open %s"), curr_pack);
 152         }
 153
 154         threads_active = 1;
 155 }
 156
 157 static void cleanup_thread(void)
 158 {
 159         int i;
 160         if (!threads_active)
 161                 return;
 162         threads_active = 0;
 163         pthread_mutex_destroy(&read_mutex);
 164         pthread_mutex_destroy(&counter_mutex);
 165         pthread_mutex_destroy(&work_mutex);
 166         pthread_mutex_destroy(&type_cas_mutex);
 167         if (show_stat)
 168                 pthread_mutex_destroy(&deepest_delta_mutex);
 169         for (i = 0; i < nr_threads; i++)
 170                 close(thread_data[i].pack_fd);
 171         pthread_key_delete(key);
 172         free(thread_data);
 173 }
 174
 175 #else
 176
 177 #define read_lock()
 178 #define read_unlock()
 179
 180 #define counter_lock()
 181 #define counter_unlock()
 182
 183 #define work_lock()
 184 #define work_unlock()
 185
 186 #define deepest_delta_lock()
 187 #define deepest_delta_unlock()
 188
 189 #define type_cas_lock()
 190 #define type_cas_unlock()
 191
 192 #endif
 193
 194
 195 static int mark_link(struct object *obj, int type, void *data)
 196 {
 197         if (!obj)
 198                 return -1;
 199
 200         if (type != OBJ_ANY && obj->type != type)
 201                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 202
 203         obj->flags |= FLAG_LINK;
 204         return 0;
 205 }
 206
 207 /* The content of each linked object must have been checked
 208    or it must be already present in the object database */
 209 static unsigned check_object(struct object *obj)
 210 {
 211         if (!obj)
 212                 return 0;
 213
 214         if (!(obj->flags & FLAG_LINK))
 215                 return 0;
 216
 217         if (!(obj->flags & FLAG_CHECKED)) {
 218                 unsigned long size;
 219                 int type = sha1_object_info(obj->sha1, &size);
 220                 if (type <= 0)
 221                         die(_("did not receive expected object %s"),
 222                               sha1_to_hex(obj->sha1));
 223                 if (type != obj->type)
 224                         die(_("object %s: expected type %s, found %s"),
 225                             sha1_to_hex(obj->sha1),
 226                             typename(obj->type), typename(type));
 227                 obj->flags |= FLAG_CHECKED;
 228                 return 1;
 229         }
 230
 231         return 0;
 232 }
 233
 234 static unsigned check_objects(void)
 235 {
 236         unsigned i, max, foreign_nr = 0;
 237
 238         max = get_max_object_index();
 239         for (i = 0; i < max; i++)
 240                 foreign_nr += check_object(get_indexed_object(i));
 241         return foreign_nr;
 242 }
 243
 244
 245 /* Discard current buffer used content. */
 246 static void flush(void)
 247 {
 248         if (input_offset) {
 249                 if (output_fd >= 0)
 250                         write_or_die(output_fd, input_buffer, input_offset);
 251                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 252                 memmove(input_buffer, input_buffer + input_offset, input_len);
 253                 input_offset = 0;
 254         }
 255 }
 256
 257 /*
 258  * Make sure at least "min" bytes are available in the buffer, and
 259  * return the pointer to the buffer.
 260  */
 261 static void *fill(int min)
 262 {
 263         if (min <= input_len)
 264                 return input_buffer + input_offset;
 265         if (min > sizeof(input_buffer))
 266                 die(Q_("cannot fill %d byte",
 267                        "cannot fill %d bytes",
 268                        min),
 269                     min);
 270         flush();
 271         do {
 272                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 273                                 sizeof(input_buffer) - input_len);
 274                 if (ret <= 0) {
 275                         if (!ret)
 276                                 die(_("early EOF"));
 277                         die_errno(_("read error on input"));
 278                 }
 279                 input_len += ret;
 280                 if (from_stdin)
 281                         display_throughput(progress, consumed_bytes + input_len);
 282         } while (input_len < min);
 283         return input_buffer;
 284 }
 285
 286 static void use(int bytes)
 287 {
 288         if (bytes > input_len)
 289                 die(_("used more bytes than were available"));
 290         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 291         input_len -= bytes;
 292         input_offset += bytes;
 293
 294         /* make sure off_t is sufficiently large not to wrap */
 295         if (signed_add_overflows(consumed_bytes, bytes))
 296                 die(_("pack too large for current definition of off_t"));
 297         consumed_bytes += bytes;
 298 }
 299
 300 static const char *open_pack_file(const char *pack_name)
 301 {
 302         if (from_stdin) {
 303                 input_fd = 0;
 304                 if (!pack_name) {
 305                         static char tmp_file[PATH_MAX];
 306                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 307                                                 "pack/tmp_pack_XXXXXX");
 308                         pack_name = xstrdup(tmp_file);
 309                 } else
 310                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 311                 if (output_fd < 0)
 312                         die_errno(_("unable to create '%s'"), pack_name);
 313                 nothread_data.pack_fd = output_fd;
 314         } else {
 315                 input_fd = open(pack_name, O_RDONLY);
 316                 if (input_fd < 0)
 317                         die_errno(_("cannot open packfile '%s'"), pack_name);
 318                 output_fd = -1;
 319                 nothread_data.pack_fd = input_fd;
 320         }
 321         git_SHA1_Init(&input_ctx);
 322         return pack_name;
 323 }
 324
 325 static void parse_pack_header(void)
 326 {
 327         struct pack_header *hdr = fill(sizeof(struct pack_header));
 328
 329         /* Header consistency check */
 330         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 331                 die(_("pack signature mismatch"));
 332         if (!pack_version_ok(hdr->hdr_version))
 333                 die(_("pack version %"PRIu32" unsupported"),
 334                         ntohl(hdr->hdr_version));
 335
 336         nr_objects = ntohl(hdr->hdr_entries);
 337         use(sizeof(struct pack_header));
 338 }
 339
 340 static NORETURN void bad_object(unsigned long offset, const char *format,
 341                        ...) __attribute__((format (printf, 2, 3)));
 342
 343 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 344 {
 345         va_list params;
 346         char buf[1024];
 347
 348         va_start(params, format);
 349         vsnprintf(buf, sizeof(buf), format, params);
 350         va_end(params);
 351         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 352 }
 353
 354 static inline struct thread_local *get_thread_data(void)
 355 {
 356 #ifndef NO_PTHREADS
 357         if (threads_active)
 358                 return pthread_getspecific(key);
 359         assert(!threads_active &&
 360                "This should only be reached when all threads are gone");
 361 #endif
 362         return &nothread_data;
 363 }
 364
 365 #ifndef NO_PTHREADS
 366 static void set_thread_data(struct thread_local *data)
 367 {
 368         if (threads_active)
 369                 pthread_setspecific(key, data);
 370 }
 371 #endif
 372
 373 static struct base_data *alloc_base_data(void)
 374 {
 375         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 376         base->ref_last = -1;
 377         base->ofs_last = -1;
 378         return base;
 379 }
 380
 381 static void free_base_data(struct base_data *c)
 382 {
 383         if (c->data) {
 384                 free(c->data);
 385                 c->data = NULL;
 386                 get_thread_data()->base_cache_used -= c->size;
 387         }
 388 }
 389
 390 static void prune_base_data(struct base_data *retain)
 391 {
 392         struct base_data *b;
 393         struct thread_local *data = get_thread_data();
 394         for (b = data->base_cache;
 395              data->base_cache_used > delta_base_cache_limit && b;
 396              b = b->child) {
 397                 if (b->data && b != retain)
 398                         free_base_data(b);
 399         }
 400 }
 401
 402 static void link_base_data(struct base_data *base, struct base_data *c)
 403 {
 404         if (base)
 405                 base->child = c;
 406         else
 407                 get_thread_data()->base_cache = c;
 408
 409         c->base = base;
 410         c->child = NULL;
 411         if (c->data)
 412                 get_thread_data()->base_cache_used += c->size;
 413         prune_base_data(c);
 414 }
 415
 416 static void unlink_base_data(struct base_data *c)
 417 {
 418         struct base_data *base = c->base;
 419         if (base)
 420                 base->child = NULL;
 421         else
 422                 get_thread_data()->base_cache = NULL;
 423         free_base_data(c);
 424 }
 425
 426 static int is_delta_type(enum object_type type)
 427 {
 428         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 429 }
 430
 431 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 432                                enum object_type type, unsigned char *sha1)
 433 {
 434         static char fixed_buf[8192];
 435         int status;
 436         git_zstream stream;
 437         void *buf;
 438         git_SHA_CTX c;
 439         char hdr[32];
 440         int hdrlen;
 441
 442         if (!is_delta_type(type)) {
 443                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 444                 git_SHA1_Init(&c);
 445                 git_SHA1_Update(&c, hdr, hdrlen);
 446         } else
 447                 sha1 = NULL;
 448         if (type == OBJ_BLOB && size > big_file_threshold)
 449                 buf = fixed_buf;
 450         else
 451                 buf = xmallocz(size);
 452
 453         memset(&stream, 0, sizeof(stream));
 454         git_inflate_init(&stream);
 455         stream.next_out = buf;
 456         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 457
 458         do {
 459                 unsigned char *last_out = stream.next_out;
 460                 stream.next_in = fill(1);
 461                 stream.avail_in = input_len;
 462                 status = git_inflate(&stream, 0);
 463                 use(input_len - stream.avail_in);
 464                 if (sha1)
 465                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 466                 if (buf == fixed_buf) {
 467                         stream.next_out = buf;
 468                         stream.avail_out = sizeof(fixed_buf);
 469                 }
 470         } while (status == Z_OK);
 471         if (stream.total_out != size || status != Z_STREAM_END)
 472                 bad_object(offset, _("inflate returned %d"), status);
 473         git_inflate_end(&stream);
 474         if (sha1)
 475                 git_SHA1_Final(sha1, &c);
 476         return buf == fixed_buf ? NULL : buf;
 477 }
 478
 479 static void *unpack_raw_entry(struct object_entry *obj,
 480                               off_t *ofs_offset,
 481                               unsigned char *ref_sha1,
 482                               unsigned char *sha1)
 483 {
 484         unsigned char *p;
 485         unsigned long size, c;
 486         off_t base_offset;
 487         unsigned shift;
 488         void *data;
 489
 490         obj->idx.offset = consumed_bytes;
 491         input_crc32 = crc32(0, NULL, 0);
 492
 493         p = fill(1);
 494         c = *p;
 495         use(1);
 496         obj->type = (c >> 4) & 7;
 497         size = (c & 15);
 498         shift = 4;
 499         while (c & 0x80) {
 500                 p = fill(1);
 501                 c = *p;
 502                 use(1);
 503                 size += (c & 0x7f) << shift;
 504                 shift += 7;
 505         }
 506         obj->size = size;
 507
 508         switch (obj->type) {
 509         case OBJ_REF_DELTA:
 510                 hashcpy(ref_sha1, fill(20));
 511                 use(20);
 512                 break;
 513         case OBJ_OFS_DELTA:
 514                 p = fill(1);
 515                 c = *p;
 516                 use(1);
 517                 base_offset = c & 127;
 518                 while (c & 128) {
 519                         base_offset += 1;
 520                         if (!base_offset || MSB(base_offset, 7))
 521                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 522                         p = fill(1);
 523                         c = *p;
 524                         use(1);
 525                         base_offset = (base_offset << 7) + (c & 127);
 526                 }
 527                 *ofs_offset = obj->idx.offset - base_offset;
 528                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 529                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 530                 break;
 531         case OBJ_COMMIT:
 532         case OBJ_TREE:
 533         case OBJ_BLOB:
 534         case OBJ_TAG:
 535                 break;
 536         default:
 537                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 538         }
 539         obj->hdr_size = consumed_bytes - obj->idx.offset;
 540
 541         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 542         obj->idx.crc32 = input_crc32;
 543         return data;
 544 }
 545
 546 static void *unpack_data(struct object_entry *obj,
 547                          int (*consume)(const unsigned char *, unsigned long, void *),
 548                          void *cb_data)
 549 {
 550         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 551         unsigned long len = obj[1].idx.offset - from;
 552         unsigned char *data, *inbuf;
 553         git_zstream stream;
 554         int status;
 555
 556         data = xmallocz(consume ? 64*1024 : obj->size);
 557         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 558
 559         memset(&stream, 0, sizeof(stream));
 560         git_inflate_init(&stream);
 561         stream.next_out = data;
 562         stream.avail_out = consume ? 64*1024 : obj->size;
 563
 564         do {
 565                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 566                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 567                 if (n < 0)
 568                         die_errno(_("cannot pread pack file"));
 569                 if (!n)
 570                         die(Q_("premature end of pack file, %lu byte missing",
 571                                "premature end of pack file, %lu bytes missing",
 572                                len),
 573                             len);
 574                 from += n;
 575                 len -= n;
 576                 stream.next_in = inbuf;
 577                 stream.avail_in = n;
 578                 if (!consume)
 579                         status = git_inflate(&stream, 0);
 580                 else {
 581                         do {
 582                                 status = git_inflate(&stream, 0);
 583                                 if (consume(data, stream.next_out - data, cb_data)) {
 584                                         free(inbuf);
 585                                         free(data);
 586                                         return NULL;
 587                                 }
 588                                 stream.next_out = data;
 589                                 stream.avail_out = 64*1024;
 590                         } while (status == Z_OK && stream.avail_in);
 591                 }
 592         } while (len && status == Z_OK && !stream.avail_in);
 593
 594         /* This has been inflated OK when first encountered, so... */
 595         if (status != Z_STREAM_END || stream.total_out != obj->size)
 596                 die(_("serious inflate inconsistency"));
 597
 598         git_inflate_end(&stream);
 599         free(inbuf);
 600         if (consume) {
 601                 free(data);
 602                 data = NULL;
 603         }
 604         return data;
 605 }
 606
 607 static void *get_data_from_pack(struct object_entry *obj)
 608 {
 609         return unpack_data(obj, NULL, NULL);
 610 }
 611
 612 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 613                                    enum object_type type1,
 614                                    enum object_type type2)
 615 {
 616         int cmp = type1 - type2;
 617         if (cmp)
 618                 return cmp;
 619         return offset1 - offset2;
 620 }
 621
 622 static int find_ofs_delta(const off_t offset, enum object_type type)
 623 {
 624         int first = 0, last = nr_ofs_deltas;
 625
 626         while (first < last) {
 627                 int next = (first + last) / 2;
 628                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 629                 int cmp;
 630
 631                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 632                                               type, objects[delta->obj_no].type);
 633                 if (!cmp)
 634                         return next;
 635                 if (cmp < 0) {
 636                         last = next;
 637                         continue;
 638                 }
 639                 first = next+1;
 640         }
 641         return -first-1;
 642 }
 643
 644 static void find_ofs_delta_children(off_t offset,
 645                                     int *first_index, int *last_index,
 646                                     enum object_type type)
 647 {
 648         int first = find_ofs_delta(offset, type);
 649         int last = first;
 650         int end = nr_ofs_deltas - 1;
 651
 652         if (first < 0) {
 653                 *first_index = 0;
 654                 *last_index = -1;
 655                 return;
 656         }
 657         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 658                 --first;
 659         while (last < end && ofs_deltas[last + 1].offset == offset)
 660                 ++last;
 661         *first_index = first;
 662         *last_index = last;
 663 }
 664
 665 static int compare_ref_delta_bases(const unsigned char *sha1,
 666                                    const unsigned char *sha2,
 667                                    enum object_type type1,
 668                                    enum object_type type2)
 669 {
 670         int cmp = type1 - type2;
 671         if (cmp)
 672                 return cmp;
 673         return hashcmp(sha1, sha2);
 674 }
 675
 676 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
 677 {
 678         int first = 0, last = nr_ref_deltas;
 679
 680         while (first < last) {
 681                 int next = (first + last) / 2;
 682                 struct ref_delta_entry *delta = &ref_deltas[next];
 683                 int cmp;
 684
 685                 cmp = compare_ref_delta_bases(sha1, delta->sha1,
 686                                               type, objects[delta->obj_no].type);
 687                 if (!cmp)
 688                         return next;
 689                 if (cmp < 0) {
 690                         last = next;
 691                         continue;
 692                 }
 693                 first = next+1;
 694         }
 695         return -first-1;
 696 }
 697
 698 static void find_ref_delta_children(const unsigned char *sha1,
 699                                     int *first_index, int *last_index,
 700                                     enum object_type type)
 701 {
 702         int first = find_ref_delta(sha1, type);
 703         int last = first;
 704         int end = nr_ref_deltas - 1;
 705
 706         if (first < 0) {
 707                 *first_index = 0;
 708                 *last_index = -1;
 709                 return;
 710         }
 711         while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
 712                 --first;
 713         while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
 714                 ++last;
 715         *first_index = first;
 716         *last_index = last;
 717 }
 718
 719 struct compare_data {
 720         struct object_entry *entry;
 721         struct git_istream *st;
 722         unsigned char *buf;
 723         unsigned long buf_size;
 724 };
 725
 726 static int compare_objects(const unsigned char *buf, unsigned long size,
 727                            void *cb_data)
 728 {
 729         struct compare_data *data = cb_data;
 730
 731         if (data->buf_size < size) {
 732                 free(data->buf);
 733                 data->buf = xmalloc(size);
 734                 data->buf_size = size;
 735         }
 736
 737         while (size) {
 738                 ssize_t len = read_istream(data->st, data->buf, size);
 739                 if (len == 0)
 740                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 741                             sha1_to_hex(data->entry->idx.sha1));
 742                 if (len < 0)
 743                         die(_("unable to read %s"),
 744                             sha1_to_hex(data->entry->idx.sha1));
 745                 if (memcmp(buf, data->buf, len))
 746                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 747                             sha1_to_hex(data->entry->idx.sha1));
 748                 size -= len;
 749                 buf += len;
 750         }
 751         return 0;
 752 }
 753
 754 static int check_collison(struct object_entry *entry)
 755 {
 756         struct compare_data data;
 757         enum object_type type;
 758         unsigned long size;
 759
 760         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 761                 return -1;
 762
 763         memset(&data, 0, sizeof(data));
 764         data.entry = entry;
 765         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 766         if (!data.st)
 767                 return -1;
 768         if (size != entry->size || type != entry->type)
 769                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 770                     sha1_to_hex(entry->idx.sha1));
 771         unpack_data(entry, compare_objects, &data);
 772         close_istream(data.st);
 773         free(data.buf);
 774         return 0;
 775 }
 776
 777 static void sha1_object(const void *data, struct object_entry *obj_entry,
 778                         unsigned long size, enum object_type type,
 779                         const unsigned char *sha1)
 780 {
 781         void *new_data = NULL;
 782         int collision_test_needed;
 783
 784         assert(data || obj_entry);
 785
 786         read_lock();
 787         collision_test_needed = has_sha1_file(sha1);
 788         read_unlock();
 789
 790         if (collision_test_needed && !data) {
 791                 read_lock();
 792                 if (!check_collison(obj_entry))
 793                         collision_test_needed = 0;
 794                 read_unlock();
 795         }
 796         if (collision_test_needed) {
 797                 void *has_data;
 798                 enum object_type has_type;
 799                 unsigned long has_size;
 800                 read_lock();
 801                 has_type = sha1_object_info(sha1, &has_size);
 802                 if (has_type != type || has_size != size)
 803                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 804                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 805                 read_unlock();
 806                 if (!data)
 807                         data = new_data = get_data_from_pack(obj_entry);
 808                 if (!has_data)
 809                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 810                 if (size != has_size || type != has_type ||
 811                     memcmp(data, has_data, size) != 0)
 812                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 813                 free(has_data);
 814         }
 815
 816         if (strict) {
 817                 read_lock();
 818                 if (type == OBJ_BLOB) {
 819                         struct blob *blob = lookup_blob(sha1);
 820                         if (blob)
 821                                 blob->object.flags |= FLAG_CHECKED;
 822                         else
 823                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 824                 } else {
 825                         struct object *obj;
 826                         int eaten;
 827                         void *buf = (void *) data;
 828
 829                         assert(data && "data can only be NULL for large _blobs_");
 830
 831                         /*
 832                          * we do not need to free the memory here, as the
 833                          * buf is deleted by the caller.
 834                          */
 835                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 836                         if (!obj)
 837                                 die(_("invalid %s"), typename(type));
 838                         if (do_fsck_object &&
 839                             fsck_object(obj, buf, size, 1,
 840                                     fsck_error_function))
 841                                 die(_("Error in object"));
 842                         if (fsck_walk(obj, mark_link, NULL))
 843                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 844
 845                         if (obj->type == OBJ_TREE) {
 846                                 struct tree *item = (struct tree *) obj;
 847                                 item->buffer = NULL;
 848                                 obj->parsed = 0;
 849                         }
 850                         if (obj->type == OBJ_COMMIT) {
 851                                 struct commit *commit = (struct commit *) obj;
 852                                 if (detach_commit_buffer(commit, NULL) != data)
 853                                         die("BUG: parse_object_buffer transmogrified our buffer");
 854                         }
 855                         obj->flags |= FLAG_CHECKED;
 856                 }
 857                 read_unlock();
 858         }
 859
 860         free(new_data);
 861 }
 862
 863 /*
 864  * This function is part of find_unresolved_deltas(). There are two
 865  * walkers going in the opposite ways.
 866  *
 867  * The first one in find_unresolved_deltas() traverses down from
 868  * parent node to children, deflating nodes along the way. However,
 869  * memory for deflated nodes is limited by delta_base_cache_limit, so
 870  * at some point parent node's deflated content may be freed.
 871  *
 872  * The second walker is this function, which goes from current node up
 873  * to top parent if necessary to deflate the node. In normal
 874  * situation, its parent node would be already deflated, so it just
 875  * needs to apply delta.
 876  *
 877  * In the worst case scenario, parent node is no longer deflated because
 878  * we're running out of delta_base_cache_limit; we need to re-deflate
 879  * parents, possibly up to the top base.
 880  *
 881  * All deflated objects here are subject to be freed if we exceed
 882  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 883  * just need to make sure the last node is not freed.
 884  */
 885 static void *get_base_data(struct base_data *c)
 886 {
 887         if (!c->data) {
 888                 struct object_entry *obj = c->obj;
 889                 struct base_data **delta = NULL;
 890                 int delta_nr = 0, delta_alloc = 0;
 891
 892                 while (is_delta_type(c->obj->type) && !c->data) {
 893                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 894                         delta[delta_nr++] = c;
 895                         c = c->base;
 896                 }
 897                 if (!delta_nr) {
 898                         c->data = get_data_from_pack(obj);
 899                         c->size = obj->size;
 900                         get_thread_data()->base_cache_used += c->size;
 901                         prune_base_data(c);
 902                 }
 903                 for (; delta_nr > 0; delta_nr--) {
 904                         void *base, *raw;
 905                         c = delta[delta_nr - 1];
 906                         obj = c->obj;
 907                         base = get_base_data(c->base);
 908                         raw = get_data_from_pack(obj);
 909                         c->data = patch_delta(
 910                                 base, c->base->size,
 911                                 raw, obj->size,
 912                                 &c->size);
 913                         free(raw);
 914                         if (!c->data)
 915                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 916                         get_thread_data()->base_cache_used += c->size;
 917                         prune_base_data(c);
 918                 }
 919                 free(delta);
 920         }
 921         return c->data;
 922 }
 923
 924 static void resolve_delta(struct object_entry *delta_obj,
 925                           struct base_data *base, struct base_data *result)
 926 {
 927         void *base_data, *delta_data;
 928
 929         if (show_stat) {
 930                 int i = delta_obj - objects;
 931                 int j = base->obj - objects;
 932                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 933                 deepest_delta_lock();
 934                 if (deepest_delta < obj_stat[i].delta_depth)
 935                         deepest_delta = obj_stat[i].delta_depth;
 936                 deepest_delta_unlock();
 937                 obj_stat[i].base_object_no = j;
 938         }
 939         delta_data = get_data_from_pack(delta_obj);
 940         base_data = get_base_data(base);
 941         result->obj = delta_obj;
 942         result->data = patch_delta(base_data, base->size,
 943                                    delta_data, delta_obj->size, &result->size);
 944         free(delta_data);
 945         if (!result->data)
 946                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 947         hash_sha1_file(result->data, result->size,
 948                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 949         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 950                     delta_obj->idx.sha1);
 951         counter_lock();
 952         nr_resolved_deltas++;
 953         counter_unlock();
 954 }
 955
 956 /*
 957  * Standard boolean compare-and-swap: atomically check whether "*type" is
 958  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 959  * and return false.
 960  */
 961 static int compare_and_swap_type(signed char *type,
 962                                  enum object_type want,
 963                                  enum object_type set)
 964 {
 965         enum object_type old;
 966
 967         type_cas_lock();
 968         old = *type;
 969         if (old == want)
 970                 *type = set;
 971         type_cas_unlock();
 972
 973         return old == want;
 974 }
 975
 976 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 977                                                   struct base_data *prev_base)
 978 {
 979         if (base->ref_last == -1 && base->ofs_last == -1) {
 980                 find_ref_delta_children(base->obj->idx.sha1,
 981                                         &base->ref_first, &base->ref_last,
 982                                         OBJ_REF_DELTA);
 983
 984                 find_ofs_delta_children(base->obj->idx.offset,
 985                                         &base->ofs_first, &base->ofs_last,
 986                                         OBJ_OFS_DELTA);
 987
 988                 if (base->ref_last == -1 && base->ofs_last == -1) {
 989                         free(base->data);
 990                         return NULL;
 991                 }
 992
 993                 link_base_data(prev_base, base);
 994         }
 995
 996         if (base->ref_first <= base->ref_last) {
 997                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
 998                 struct base_data *result = alloc_base_data();
 999
1000                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1001                                            base->obj->real_type))
1002                         die("BUG: child->real_type != OBJ_REF_DELTA");
1003
1004                 resolve_delta(child, base, result);
1005                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1006                         free_base_data(base);
1007
1008                 base->ref_first++;
1009                 return result;
1010         }
1011
1012         if (base->ofs_first <= base->ofs_last) {
1013                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1014                 struct base_data *result = alloc_base_data();
1015
1016                 assert(child->real_type == OBJ_OFS_DELTA);
1017                 child->real_type = base->obj->real_type;
1018                 resolve_delta(child, base, result);
1019                 if (base->ofs_first == base->ofs_last)
1020                         free_base_data(base);
1021
1022                 base->ofs_first++;
1023                 return result;
1024         }
1025
1026         unlink_base_data(base);
1027         return NULL;
1028 }
1029
1030 static void find_unresolved_deltas(struct base_data *base)
1031 {
1032         struct base_data *new_base, *prev_base = NULL;
1033         for (;;) {
1034                 new_base = find_unresolved_deltas_1(base, prev_base);
1035
1036                 if (new_base) {
1037                         prev_base = base;
1038                         base = new_base;
1039                 } else {
1040                         free(base);
1041                         base = prev_base;
1042                         if (!base)
1043                                 return;
1044                         prev_base = base->base;
1045                 }
1046         }
1047 }
1048
1049 static int compare_ofs_delta_entry(const void *a, const void *b)
1050 {
1051         const struct ofs_delta_entry *delta_a = a;
1052         const struct ofs_delta_entry *delta_b = b;
1053
1054         return delta_a->offset - delta_b->offset;
1055 }
1056
1057 static int compare_ref_delta_entry(const void *a, const void *b)
1058 {
1059         const struct ref_delta_entry *delta_a = a;
1060         const struct ref_delta_entry *delta_b = b;
1061
1062         return hashcmp(delta_a->sha1, delta_b->sha1);
1063 }
1064
1065 static void resolve_base(struct object_entry *obj)
1066 {
1067         struct base_data *base_obj = alloc_base_data();
1068         base_obj->obj = obj;
1069         base_obj->data = NULL;
1070         find_unresolved_deltas(base_obj);
1071 }
1072
1073 #ifndef NO_PTHREADS
1074 static void *threaded_second_pass(void *data)
1075 {
1076         set_thread_data(data);
1077         for (;;) {
1078                 int i;
1079                 counter_lock();
1080                 display_progress(progress, nr_resolved_deltas);
1081                 counter_unlock();
1082                 work_lock();
1083                 while (nr_dispatched < nr_objects &&
1084                        is_delta_type(objects[nr_dispatched].type))
1085                         nr_dispatched++;
1086                 if (nr_dispatched >= nr_objects) {
1087                         work_unlock();
1088                         break;
1089                 }
1090                 i = nr_dispatched++;
1091                 work_unlock();
1092
1093                 resolve_base(&objects[i]);
1094         }
1095         return NULL;
1096 }
1097 #endif
1098
1099 /*
1100  * First pass:
1101  * - find locations of all objects;
1102  * - calculate SHA1 of all non-delta objects;
1103  * - remember base (SHA1 or offset) for all deltas.
1104  */
1105 static void parse_pack_objects(unsigned char *sha1)
1106 {
1107         int i, nr_delays = 0;
1108         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1109         unsigned char ref_delta_sha1[20];
1110         struct stat st;
1111
1112         if (verbose)
1113                 progress = start_progress(
1114                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1115                                 nr_objects);
1116         for (i = 0; i < nr_objects; i++) {
1117                 struct object_entry *obj = &objects[i];
1118                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1119                                               ref_delta_sha1, obj->idx.sha1);
1120                 obj->real_type = obj->type;
1121                 if (obj->type == OBJ_OFS_DELTA) {
1122                         nr_ofs_deltas++;
1123                         ofs_delta->obj_no = i;
1124                         ofs_delta++;
1125                 } else if (obj->type == OBJ_REF_DELTA) {
1126                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1127                         hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1);
1128                         ref_deltas[nr_ref_deltas].obj_no = i;
1129                         nr_ref_deltas++;
1130                 } else if (!data) {
1131                         /* large blobs, check later */
1132                         obj->real_type = OBJ_BAD;
1133                         nr_delays++;
1134                 } else
1135                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1136                 free(data);
1137                 display_progress(progress, i+1);
1138         }
1139         objects[i].idx.offset = consumed_bytes;
1140         stop_progress(&progress);
1141
1142         /* Check pack integrity */
1143         flush();
1144         git_SHA1_Final(sha1, &input_ctx);
1145         if (hashcmp(fill(20), sha1))
1146                 die(_("pack is corrupted (SHA1 mismatch)"));
1147         use(20);
1148
1149         /* If input_fd is a file, we should have reached its end now. */
1150         if (fstat(input_fd, &st))
1151                 die_errno(_("cannot fstat packfile"));
1152         if (S_ISREG(st.st_mode) &&
1153                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1154                 die(_("pack has junk at the end"));
1155
1156         for (i = 0; i < nr_objects; i++) {
1157                 struct object_entry *obj = &objects[i];
1158                 if (obj->real_type != OBJ_BAD)
1159                         continue;
1160                 obj->real_type = obj->type;
1161                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1162                 nr_delays--;
1163         }
1164         if (nr_delays)
1165                 die(_("confusion beyond insanity in parse_pack_objects()"));
1166 }
1167
1168 /*
1169  * Second pass:
1170  * - for all non-delta objects, look if it is used as a base for
1171  *   deltas;
1172  * - if used as a base, uncompress the object and apply all deltas,
1173  *   recursively checking if the resulting object is used as a base
1174  *   for some more deltas.
1175  */
1176 static void resolve_deltas(void)
1177 {
1178         int i;
1179
1180         if (!nr_ofs_deltas && !nr_ref_deltas)
1181                 return;
1182
1183         /* Sort deltas by base SHA1/offset for fast searching */
1184         qsort(ofs_deltas, nr_ofs_deltas, sizeof(struct ofs_delta_entry),
1185               compare_ofs_delta_entry);
1186         qsort(ref_deltas, nr_ref_deltas, sizeof(struct ref_delta_entry),
1187               compare_ref_delta_entry);
1188
1189         if (verbose)
1190                 progress = start_progress(_("Resolving deltas"),
1191                                           nr_ref_deltas + nr_ofs_deltas);
1192
1193 #ifndef NO_PTHREADS
1194         nr_dispatched = 0;
1195         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1196                 init_thread();
1197                 for (i = 0; i < nr_threads; i++) {
1198                         int ret = pthread_create(&thread_data[i].thread, NULL,
1199                                                  threaded_second_pass, thread_data + i);
1200                         if (ret)
1201                                 die(_("unable to create thread: %s"),
1202                                     strerror(ret));
1203                 }
1204                 for (i = 0; i < nr_threads; i++)
1205                         pthread_join(thread_data[i].thread, NULL);
1206                 cleanup_thread();
1207                 return;
1208         }
1209 #endif
1210
1211         for (i = 0; i < nr_objects; i++) {
1212                 struct object_entry *obj = &objects[i];
1213
1214                 if (is_delta_type(obj->type))
1215                         continue;
1216                 resolve_base(obj);
1217                 display_progress(progress, nr_resolved_deltas);
1218         }
1219 }
1220
1221 /*
1222  * Third pass:
1223  * - append objects to convert thin pack to full pack if required
1224  * - write the final 20-byte SHA-1
1225  */
1226 static void fix_unresolved_deltas(struct sha1file *f);
1227 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1228 {
1229         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1230                 stop_progress(&progress);
1231                 /* Flush remaining pack final 20-byte SHA1. */
1232                 flush();
1233                 return;
1234         }
1235
1236         if (fix_thin_pack) {
1237                 struct sha1file *f;
1238                 unsigned char read_sha1[20], tail_sha1[20];
1239                 struct strbuf msg = STRBUF_INIT;
1240                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1241                 int nr_objects_initial = nr_objects;
1242                 if (nr_unresolved <= 0)
1243                         die(_("confusion beyond insanity"));
1244                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1245                 memset(objects + nr_objects + 1, 0,
1246                        nr_unresolved * sizeof(*objects));
1247                 f = sha1fd(output_fd, curr_pack);
1248                 fix_unresolved_deltas(f);
1249                 strbuf_addf(&msg, _("completed with %d local objects"),
1250                             nr_objects - nr_objects_initial);
1251                 stop_progress_msg(&progress, msg.buf);
1252                 strbuf_release(&msg);
1253                 sha1close(f, tail_sha1, 0);
1254                 hashcpy(read_sha1, pack_sha1);
1255                 fixup_pack_header_footer(output_fd, pack_sha1,
1256                                          curr_pack, nr_objects,
1257                                          read_sha1, consumed_bytes-20);
1258                 if (hashcmp(read_sha1, tail_sha1) != 0)
1259                         die(_("Unexpected tail checksum for %s "
1260                               "(disk corruption?)"), curr_pack);
1261         }
1262         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1263                 die(Q_("pack has %d unresolved delta",
1264                        "pack has %d unresolved deltas",
1265                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1266                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1267 }
1268
1269 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1270 {
1271         git_zstream stream;
1272         int status;
1273         unsigned char outbuf[4096];
1274
1275         memset(&stream, 0, sizeof(stream));
1276         git_deflate_init(&stream, zlib_compression_level);
1277         stream.next_in = in;
1278         stream.avail_in = size;
1279
1280         do {
1281                 stream.next_out = outbuf;
1282                 stream.avail_out = sizeof(outbuf);
1283                 status = git_deflate(&stream, Z_FINISH);
1284                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1285         } while (status == Z_OK);
1286
1287         if (status != Z_STREAM_END)
1288                 die(_("unable to deflate appended object (%d)"), status);
1289         size = stream.total_out;
1290         git_deflate_end(&stream);
1291         return size;
1292 }
1293
1294 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1295                                const unsigned char *sha1, void *buf,
1296                                unsigned long size, enum object_type type)
1297 {
1298         struct object_entry *obj = &objects[nr_objects++];
1299         unsigned char header[10];
1300         unsigned long s = size;
1301         int n = 0;
1302         unsigned char c = (type << 4) | (s & 15);
1303         s >>= 4;
1304         while (s) {
1305                 header[n++] = c | 0x80;
1306                 c = s & 0x7f;
1307                 s >>= 7;
1308         }
1309         header[n++] = c;
1310         crc32_begin(f);
1311         sha1write(f, header, n);
1312         obj[0].size = size;
1313         obj[0].hdr_size = n;
1314         obj[0].type = type;
1315         obj[0].real_type = type;
1316         obj[1].idx.offset = obj[0].idx.offset + n;
1317         obj[1].idx.offset += write_compressed(f, buf, size);
1318         obj[0].idx.crc32 = crc32_end(f);
1319         sha1flush(f);
1320         hashcpy(obj->idx.sha1, sha1);
1321         return obj;
1322 }
1323
1324 static int delta_pos_compare(const void *_a, const void *_b)
1325 {
1326         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1327         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1328         return a->obj_no - b->obj_no;
1329 }
1330
1331 static void fix_unresolved_deltas(struct sha1file *f)
1332 {
1333         struct ref_delta_entry **sorted_by_pos;
1334         int i;
1335
1336         /*
1337          * Since many unresolved deltas may well be themselves base objects
1338          * for more unresolved deltas, we really want to include the
1339          * smallest number of base objects that would cover as much delta
1340          * as possible by picking the
1341          * trunc deltas first, allowing for other deltas to resolve without
1342          * additional base objects.  Since most base objects are to be found
1343          * before deltas depending on them, a good heuristic is to start
1344          * resolving deltas in the same order as their position in the pack.
1345          */
1346         sorted_by_pos = xmalloc(nr_ref_deltas * sizeof(*sorted_by_pos));
1347         for (i = 0; i < nr_ref_deltas; i++)
1348                 sorted_by_pos[i] = &ref_deltas[i];
1349         qsort(sorted_by_pos, nr_ref_deltas, sizeof(*sorted_by_pos), delta_pos_compare);
1350
1351         for (i = 0; i < nr_ref_deltas; i++) {
1352                 struct ref_delta_entry *d = sorted_by_pos[i];
1353                 enum object_type type;
1354                 struct base_data *base_obj = alloc_base_data();
1355
1356                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1357                         continue;
1358                 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1359                 if (!base_obj->data)
1360                         continue;
1361
1362                 if (check_sha1_signature(d->sha1, base_obj->data,
1363                                 base_obj->size, typename(type)))
1364                         die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1365                 base_obj->obj = append_obj_to_pack(f, d->sha1,
1366                                         base_obj->data, base_obj->size, type);
1367                 find_unresolved_deltas(base_obj);
1368                 display_progress(progress, nr_resolved_deltas);
1369         }
1370         free(sorted_by_pos);
1371 }
1372
1373 static void final(const char *final_pack_name, const char *curr_pack_name,
1374                   const char *final_index_name, const char *curr_index_name,
1375                   const char *keep_name, const char *keep_msg,
1376                   unsigned char *sha1)
1377 {
1378         const char *report = "pack";
1379         char name[PATH_MAX];
1380         int err;
1381
1382         if (!from_stdin) {
1383                 close(input_fd);
1384         } else {
1385                 fsync_or_die(output_fd, curr_pack_name);
1386                 err = close(output_fd);
1387                 if (err)
1388                         die_errno(_("error while closing pack file"));
1389         }
1390
1391         if (keep_msg) {
1392                 int keep_fd, keep_msg_len = strlen(keep_msg);
1393
1394                 if (!keep_name)
1395                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1396                 else
1397                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1398
1399                 if (keep_fd < 0) {
1400                         if (errno != EEXIST)
1401                                 die_errno(_("cannot write keep file '%s'"),
1402                                           keep_name ? keep_name : name);
1403                 } else {
1404                         if (keep_msg_len > 0) {
1405                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1406                                 write_or_die(keep_fd, "\n", 1);
1407                         }
1408                         if (close(keep_fd) != 0)
1409                                 die_errno(_("cannot close written keep file '%s'"),
1410                                           keep_name ? keep_name : name);
1411                         report = "keep";
1412                 }
1413         }
1414
1415         if (final_pack_name != curr_pack_name) {
1416                 if (!final_pack_name) {
1417                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1418                                  get_object_directory(), sha1_to_hex(sha1));
1419                         final_pack_name = name;
1420                 }
1421                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1422                         die(_("cannot store pack file"));
1423         } else if (from_stdin)
1424                 chmod(final_pack_name, 0444);
1425
1426         if (final_index_name != curr_index_name) {
1427                 if (!final_index_name) {
1428                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1429                                  get_object_directory(), sha1_to_hex(sha1));
1430                         final_index_name = name;
1431                 }
1432                 if (move_temp_to_file(curr_index_name, final_index_name))
1433                         die(_("cannot store index file"));
1434         } else
1435                 chmod(final_index_name, 0444);
1436
1437         if (!from_stdin) {
1438                 printf("%s\n", sha1_to_hex(sha1));
1439         } else {
1440                 char buf[48];
1441                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1442                                    report, sha1_to_hex(sha1));
1443                 write_or_die(1, buf, len);
1444
1445                 /*
1446                  * Let's just mimic git-unpack-objects here and write
1447                  * the last part of the input buffer to stdout.
1448                  */
1449                 while (input_len) {
1450                         err = xwrite(1, input_buffer + input_offset, input_len);
1451                         if (err <= 0)
1452                                 break;
1453                         input_len -= err;
1454                         input_offset += err;
1455                 }
1456         }
1457 }
1458
1459 static int git_index_pack_config(const char *k, const char *v, void *cb)
1460 {
1461         struct pack_idx_option *opts = cb;
1462
1463         if (!strcmp(k, "pack.indexversion")) {
1464                 opts->version = git_config_int(k, v);
1465                 if (opts->version > 2)
1466                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1467                 return 0;
1468         }
1469         if (!strcmp(k, "pack.threads")) {
1470                 nr_threads = git_config_int(k, v);
1471                 if (nr_threads < 0)
1472                         die(_("invalid number of threads specified (%d)"),
1473                             nr_threads);
1474 #ifdef NO_PTHREADS
1475                 if (nr_threads != 1)
1476                         warning(_("no threads support, ignoring %s"), k);
1477                 nr_threads = 1;
1478 #endif
1479                 return 0;
1480         }
1481         return git_default_config(k, v, cb);
1482 }
1483
1484 static int cmp_uint32(const void *a_, const void *b_)
1485 {
1486         uint32_t a = *((uint32_t *)a_);
1487         uint32_t b = *((uint32_t *)b_);
1488
1489         return (a < b) ? -1 : (a != b);
1490 }
1491
1492 static void read_v2_anomalous_offsets(struct packed_git *p,
1493                                       struct pack_idx_option *opts)
1494 {
1495         const uint32_t *idx1, *idx2;
1496         uint32_t i;
1497
1498         /* The address of the 4-byte offset table */
1499         idx1 = (((const uint32_t *)p->index_data)
1500                 + 2 /* 8-byte header */
1501                 + 256 /* fan out */
1502                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1503                 + p->num_objects /* CRC32 table */
1504                 );
1505
1506         /* The address of the 8-byte offset table */
1507         idx2 = idx1 + p->num_objects;
1508
1509         for (i = 0; i < p->num_objects; i++) {
1510                 uint32_t off = ntohl(idx1[i]);
1511                 if (!(off & 0x80000000))
1512                         continue;
1513                 off = off & 0x7fffffff;
1514                 if (idx2[off * 2])
1515                         continue;
1516                 /*
1517                  * The real offset is ntohl(idx2[off * 2]) in high 4
1518                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1519                  * octets.  But idx2[off * 2] is Zero!!!
1520                  */
1521                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1522                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1523         }
1524
1525         if (1 < opts->anomaly_nr)
1526                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1527 }
1528
1529 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1530 {
1531         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1532
1533         if (!p)
1534                 die(_("Cannot open existing pack file '%s'"), pack_name);
1535         if (open_pack_index(p))
1536                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1537
1538         /* Read the attributes from the existing idx file */
1539         opts->version = p->index_version;
1540
1541         if (opts->version == 2)
1542                 read_v2_anomalous_offsets(p, opts);
1543
1544         /*
1545          * Get rid of the idx file as we do not need it anymore.
1546          * NEEDSWORK: extract this bit from free_pack_by_name() in
1547          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1548          * know we haven't installed this pack (hence we never have
1549          * read anything from it).
1550          */
1551         close_pack_index(p);
1552         free(p);
1553 }
1554
1555 static void show_pack_info(int stat_only)
1556 {
1557         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1558         unsigned long *chain_histogram = NULL;
1559
1560         if (deepest_delta)
1561                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1562
1563         for (i = 0; i < nr_objects; i++) {
1564                 struct object_entry *obj = &objects[i];
1565
1566                 if (is_delta_type(obj->type))
1567                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1568                 if (stat_only)
1569                         continue;
1570                 printf("%s %-6s %lu %lu %"PRIuMAX,
1571                        sha1_to_hex(obj->idx.sha1),
1572                        typename(obj->real_type), obj->size,
1573                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1574                        (uintmax_t)obj->idx.offset);
1575                 if (is_delta_type(obj->type)) {
1576                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1577                         printf(" %u %s", obj_stat[i].delta_depth, sha1_to_hex(bobj->idx.sha1));
1578                 }
1579                 putchar('\n');
1580         }
1581
1582         if (baseobjects)
1583                 printf_ln(Q_("non delta: %d object",
1584                              "non delta: %d objects",
1585                              baseobjects),
1586                           baseobjects);
1587         for (i = 0; i < deepest_delta; i++) {
1588                 if (!chain_histogram[i])
1589                         continue;
1590                 printf_ln(Q_("chain length = %d: %lu object",
1591                              "chain length = %d: %lu objects",
1592                              chain_histogram[i]),
1593                           i + 1,
1594                           chain_histogram[i]);
1595         }
1596 }
1597
1598 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1599 {
1600         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1601         const char *curr_index;
1602         const char *index_name = NULL, *pack_name = NULL;
1603         const char *keep_name = NULL, *keep_msg = NULL;
1604         struct strbuf index_name_buf = STRBUF_INIT,
1605                       keep_name_buf = STRBUF_INIT;
1606         struct pack_idx_entry **idx_objects;
1607         struct pack_idx_option opts;
1608         unsigned char pack_sha1[20];
1609         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1610
1611         if (argc == 2 && !strcmp(argv[1], "-h"))
1612                 usage(index_pack_usage);
1613
1614         check_replace_refs = 0;
1615
1616         reset_pack_idx_option(&opts);
1617         git_config(git_index_pack_config, &opts);
1618         if (prefix && chdir(prefix))
1619                 die(_("Cannot come back to cwd"));
1620
1621         for (i = 1; i < argc; i++) {
1622                 const char *arg = argv[i];
1623
1624                 if (*arg == '-') {
1625                         if (!strcmp(arg, "--stdin")) {
1626                                 from_stdin = 1;
1627                         } else if (!strcmp(arg, "--fix-thin")) {
1628                                 fix_thin_pack = 1;
1629                         } else if (!strcmp(arg, "--strict")) {
1630                                 strict = 1;
1631                                 do_fsck_object = 1;
1632                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1633                                 strict = 1;
1634                                 check_self_contained_and_connected = 1;
1635                         } else if (!strcmp(arg, "--verify")) {
1636                                 verify = 1;
1637                         } else if (!strcmp(arg, "--verify-stat")) {
1638                                 verify = 1;
1639                                 show_stat = 1;
1640                         } else if (!strcmp(arg, "--verify-stat-only")) {
1641                                 verify = 1;
1642                                 show_stat = 1;
1643                                 stat_only = 1;
1644                         } else if (!strcmp(arg, "--keep")) {
1645                                 keep_msg = "";
1646                         } else if (starts_with(arg, "--keep=")) {
1647                                 keep_msg = arg + 7;
1648                         } else if (starts_with(arg, "--threads=")) {
1649                                 char *end;
1650                                 nr_threads = strtoul(arg+10, &end, 0);
1651                                 if (!arg[10] || *end || nr_threads < 0)
1652                                         usage(index_pack_usage);
1653 #ifdef NO_PTHREADS
1654                                 if (nr_threads != 1)
1655                                         warning(_("no threads support, "
1656                                                   "ignoring %s"), arg);
1657                                 nr_threads = 1;
1658 #endif
1659                         } else if (starts_with(arg, "--pack_header=")) {
1660                                 struct pack_header *hdr;
1661                                 char *c;
1662
1663                                 hdr = (struct pack_header *)input_buffer;
1664                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1665                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1666                                 if (*c != ',')
1667                                         die(_("bad %s"), arg);
1668                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1669                                 if (*c)
1670                                         die(_("bad %s"), arg);
1671                                 input_len = sizeof(*hdr);
1672                         } else if (!strcmp(arg, "-v")) {
1673                                 verbose = 1;
1674                         } else if (!strcmp(arg, "-o")) {
1675                                 if (index_name || (i+1) >= argc)
1676                                         usage(index_pack_usage);
1677                                 index_name = argv[++i];
1678                         } else if (starts_with(arg, "--index-version=")) {
1679                                 char *c;
1680                                 opts.version = strtoul(arg + 16, &c, 10);
1681                                 if (opts.version > 2)
1682                                         die(_("bad %s"), arg);
1683                                 if (*c == ',')
1684                                         opts.off32_limit = strtoul(c+1, &c, 0);
1685                                 if (*c || opts.off32_limit & 0x80000000)
1686                                         die(_("bad %s"), arg);
1687                         } else
1688                                 usage(index_pack_usage);
1689                         continue;
1690                 }
1691
1692                 if (pack_name)
1693                         usage(index_pack_usage);
1694                 pack_name = arg;
1695         }
1696
1697         if (!pack_name && !from_stdin)
1698                 usage(index_pack_usage);
1699         if (fix_thin_pack && !from_stdin)
1700                 die(_("--fix-thin cannot be used without --stdin"));
1701         if (!index_name && pack_name) {
1702                 size_t len;
1703                 if (!strip_suffix(pack_name, ".pack", &len))
1704                         die(_("packfile name '%s' does not end with '.pack'"),
1705                             pack_name);
1706                 strbuf_add(&index_name_buf, pack_name, len);
1707                 strbuf_addstr(&index_name_buf, ".idx");
1708                 index_name = index_name_buf.buf;
1709         }
1710         if (keep_msg && !keep_name && pack_name) {
1711                 size_t len;
1712                 if (!strip_suffix(pack_name, ".pack", &len))
1713                         die(_("packfile name '%s' does not end with '.pack'"),
1714                             pack_name);
1715                 strbuf_add(&keep_name_buf, pack_name, len);
1716                 strbuf_addstr(&keep_name_buf, ".idx");
1717                 keep_name = keep_name_buf.buf;
1718         }
1719         if (verify) {
1720                 if (!index_name)
1721                         die(_("--verify with no packfile name given"));
1722                 read_idx_option(&opts, index_name);
1723                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1724         }
1725         if (strict)
1726                 opts.flags |= WRITE_IDX_STRICT;
1727
1728 #ifndef NO_PTHREADS
1729         if (!nr_threads) {
1730                 nr_threads = online_cpus();
1731                 /* An experiment showed that more threads does not mean faster */
1732                 if (nr_threads > 3)
1733                         nr_threads = 3;
1734         }
1735 #endif
1736
1737         curr_pack = open_pack_file(pack_name);
1738         parse_pack_header();
1739         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1740         if (show_stat)
1741                 obj_stat = xcalloc(nr_objects + 1, sizeof(struct object_stat));
1742         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1743         parse_pack_objects(pack_sha1);
1744         resolve_deltas();
1745         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1746         free(ofs_deltas);
1747         free(ref_deltas);
1748         if (strict)
1749                 foreign_nr = check_objects();
1750
1751         if (show_stat)
1752                 show_pack_info(stat_only);
1753
1754         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1755         for (i = 0; i < nr_objects; i++)
1756                 idx_objects[i] = &objects[i].idx;
1757         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1758         free(idx_objects);
1759
1760         if (!verify)
1761                 final(pack_name, curr_pack,
1762                       index_name, curr_index,
1763                       keep_name, keep_msg,
1764                       pack_sha1);
1765         else
1766                 close(input_fd);
1767         free(objects);
1768         strbuf_release(&index_name_buf);
1769         strbuf_release(&keep_name_buf);
1770         if (pack_name == NULL)
1771                 free((void *) curr_pack);
1772         if (index_name == NULL)
1773                 free((void *) curr_index);
1774
1775         /*
1776          * Let the caller know this pack is not self contained
1777          */
1778         if (check_self_contained_and_connected && foreign_nr)
1779                 return 1;
1780
1781         return 0;
1782 }