builtin/index-pack.c

   1 #include "builtin.h"
   2 #include "config.h"
   3 #include "delta.h"
   4 #include "pack.h"
   5 #include "csum-file.h"
   6 #include "blob.h"
   7 #include "commit.h"
   8 #include "tag.h"
   9 #include "tree.h"
  10 #include "progress.h"
  11 #include "fsck.h"
  12 #include "exec-cmd.h"
  13 #include "streaming.h"
  14 #include "thread-utils.h"
  15 #include "packfile.h"
  16 #include "object-store.h"
  17 #include "promisor-remote.h"
  18
  19 static const char index_pack_usage[] =
  20 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  21
  22 struct object_entry {
  23         struct pack_idx_entry idx;
  24         unsigned long size;
  25         unsigned char hdr_size;
  26         signed char type;
  27         signed char real_type;
  28 };
  29
  30 struct object_stat {
  31         unsigned delta_depth;
  32         int base_object_no;
  33 };
  34
  35 struct base_data {
  36         struct base_data *base;
  37         struct base_data *child;
  38         struct object_entry *obj;
  39         void *data;
  40         unsigned long size;
  41         int ref_first, ref_last;
  42         int ofs_first, ofs_last;
  43 };
  44
  45 struct thread_local {
  46         pthread_t thread;
  47         struct base_data *base_cache;
  48         size_t base_cache_used;
  49         int pack_fd;
  50 };
  51
  52 /* Remember to update object flag allocation in object.h */
  53 #define FLAG_LINK (1u<<20)
  54 #define FLAG_CHECKED (1u<<21)
  55
  56 struct ofs_delta_entry {
  57         off_t offset;
  58         int obj_no;
  59 };
  60
  61 struct ref_delta_entry {
  62         struct object_id oid;
  63         int obj_no;
  64 };
  65
  66 static struct object_entry *objects;
  67 static struct object_stat *obj_stat;
  68 static struct ofs_delta_entry *ofs_deltas;
  69 static struct ref_delta_entry *ref_deltas;
  70 static struct thread_local nothread_data;
  71 static int nr_objects;
  72 static int nr_ofs_deltas;
  73 static int nr_ref_deltas;
  74 static int ref_deltas_alloc;
  75 static int nr_resolved_deltas;
  76 static int nr_threads;
  77
  78 static int from_stdin;
  79 static int strict;
  80 static int do_fsck_object;
  81 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  82 static int verbose;
  83 static int show_resolving_progress;
  84 static int show_stat;
  85 static int check_self_contained_and_connected;
  86
  87 static struct progress *progress;
  88
  89 /* We always read in 4kB chunks. */
  90 static unsigned char input_buffer[4096];
  91 static unsigned int input_offset, input_len;
  92 static off_t consumed_bytes;
  93 static off_t max_input_size;
  94 static unsigned deepest_delta;
  95 static git_hash_ctx input_ctx;
  96 static uint32_t input_crc32;
  97 static int input_fd, output_fd;
  98 static const char *curr_pack;
  99
 100 static struct thread_local *thread_data;
 101 static int nr_dispatched;
 102 static int threads_active;
 103
 104 static pthread_mutex_t read_mutex;
 105 #define read_lock()             lock_mutex(&read_mutex)
 106 #define read_unlock()           unlock_mutex(&read_mutex)
 107
 108 static pthread_mutex_t counter_mutex;
 109 #define counter_lock()          lock_mutex(&counter_mutex)
 110 #define counter_unlock()        unlock_mutex(&counter_mutex)
 111
 112 static pthread_mutex_t work_mutex;
 113 #define work_lock()             lock_mutex(&work_mutex)
 114 #define work_unlock()           unlock_mutex(&work_mutex)
 115
 116 static pthread_mutex_t deepest_delta_mutex;
 117 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 118 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 119
 120 static pthread_mutex_t type_cas_mutex;
 121 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 122 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 123
 124 static pthread_key_t key;
 125
 126 static inline void lock_mutex(pthread_mutex_t *mutex)
 127 {
 128         if (threads_active)
 129                 pthread_mutex_lock(mutex);
 130 }
 131
 132 static inline void unlock_mutex(pthread_mutex_t *mutex)
 133 {
 134         if (threads_active)
 135                 pthread_mutex_unlock(mutex);
 136 }
 137
 138 /*
 139  * Mutex and conditional variable can't be statically-initialized on Windows.
 140  */
 141 static void init_thread(void)
 142 {
 143         int i;
 144         init_recursive_mutex(&read_mutex);
 145         pthread_mutex_init(&counter_mutex, NULL);
 146         pthread_mutex_init(&work_mutex, NULL);
 147         pthread_mutex_init(&type_cas_mutex, NULL);
 148         if (show_stat)
 149                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 150         pthread_key_create(&key, NULL);
 151         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 152         for (i = 0; i < nr_threads; i++) {
 153                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 154                 if (thread_data[i].pack_fd == -1)
 155                         die_errno(_("unable to open %s"), curr_pack);
 156         }
 157
 158         threads_active = 1;
 159 }
 160
 161 static void cleanup_thread(void)
 162 {
 163         int i;
 164         if (!threads_active)
 165                 return;
 166         threads_active = 0;
 167         pthread_mutex_destroy(&read_mutex);
 168         pthread_mutex_destroy(&counter_mutex);
 169         pthread_mutex_destroy(&work_mutex);
 170         pthread_mutex_destroy(&type_cas_mutex);
 171         if (show_stat)
 172                 pthread_mutex_destroy(&deepest_delta_mutex);
 173         for (i = 0; i < nr_threads; i++)
 174                 close(thread_data[i].pack_fd);
 175         pthread_key_delete(key);
 176         free(thread_data);
 177 }
 178
 179 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 180 {
 181         if (!obj)
 182                 return -1;
 183
 184         if (type != OBJ_ANY && obj->type != type)
 185                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 186
 187         obj->flags |= FLAG_LINK;
 188         return 0;
 189 }
 190
 191 /* The content of each linked object must have been checked
 192    or it must be already present in the object database */
 193 static unsigned check_object(struct object *obj)
 194 {
 195         if (!obj)
 196                 return 0;
 197
 198         if (!(obj->flags & FLAG_LINK))
 199                 return 0;
 200
 201         if (!(obj->flags & FLAG_CHECKED)) {
 202                 unsigned long size;
 203                 int type = oid_object_info(the_repository, &obj->oid, &size);
 204                 if (type <= 0)
 205                         die(_("did not receive expected object %s"),
 206                               oid_to_hex(&obj->oid));
 207                 if (type != obj->type)
 208                         die(_("object %s: expected type %s, found %s"),
 209                             oid_to_hex(&obj->oid),
 210                             type_name(obj->type), type_name(type));
 211                 obj->flags |= FLAG_CHECKED;
 212                 return 1;
 213         }
 214
 215         return 0;
 216 }
 217
 218 static unsigned check_objects(void)
 219 {
 220         unsigned i, max, foreign_nr = 0;
 221
 222         max = get_max_object_index();
 223
 224         if (verbose)
 225                 progress = start_delayed_progress(_("Checking objects"), max);
 226
 227         for (i = 0; i < max; i++) {
 228                 foreign_nr += check_object(get_indexed_object(i));
 229                 display_progress(progress, i + 1);
 230         }
 231
 232         stop_progress(&progress);
 233         return foreign_nr;
 234 }
 235
 236
 237 /* Discard current buffer used content. */
 238 static void flush(void)
 239 {
 240         if (input_offset) {
 241                 if (output_fd >= 0)
 242                         write_or_die(output_fd, input_buffer, input_offset);
 243                 the_hash_algo->update_fn(&input_ctx, input_buffer, input_offset);
 244                 memmove(input_buffer, input_buffer + input_offset, input_len);
 245                 input_offset = 0;
 246         }
 247 }
 248
 249 /*
 250  * Make sure at least "min" bytes are available in the buffer, and
 251  * return the pointer to the buffer.
 252  */
 253 static void *fill(int min)
 254 {
 255         if (min <= input_len)
 256                 return input_buffer + input_offset;
 257         if (min > sizeof(input_buffer))
 258                 die(Q_("cannot fill %d byte",
 259                        "cannot fill %d bytes",
 260                        min),
 261                     min);
 262         flush();
 263         do {
 264                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 265                                 sizeof(input_buffer) - input_len);
 266                 if (ret <= 0) {
 267                         if (!ret)
 268                                 die(_("early EOF"));
 269                         die_errno(_("read error on input"));
 270                 }
 271                 input_len += ret;
 272                 if (from_stdin)
 273                         display_throughput(progress, consumed_bytes + input_len);
 274         } while (input_len < min);
 275         return input_buffer;
 276 }
 277
 278 static void use(int bytes)
 279 {
 280         if (bytes > input_len)
 281                 die(_("used more bytes than were available"));
 282         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 283         input_len -= bytes;
 284         input_offset += bytes;
 285
 286         /* make sure off_t is sufficiently large not to wrap */
 287         if (signed_add_overflows(consumed_bytes, bytes))
 288                 die(_("pack too large for current definition of off_t"));
 289         consumed_bytes += bytes;
 290         if (max_input_size && consumed_bytes > max_input_size)
 291                 die(_("pack exceeds maximum allowed size"));
 292 }
 293
 294 static const char *open_pack_file(const char *pack_name)
 295 {
 296         if (from_stdin) {
 297                 input_fd = 0;
 298                 if (!pack_name) {
 299                         struct strbuf tmp_file = STRBUF_INIT;
 300                         output_fd = odb_mkstemp(&tmp_file,
 301                                                 "pack/tmp_pack_XXXXXX");
 302                         pack_name = strbuf_detach(&tmp_file, NULL);
 303                 } else {
 304                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 305                         if (output_fd < 0)
 306                                 die_errno(_("unable to create '%s'"), pack_name);
 307                 }
 308                 nothread_data.pack_fd = output_fd;
 309         } else {
 310                 input_fd = open(pack_name, O_RDONLY);
 311                 if (input_fd < 0)
 312                         die_errno(_("cannot open packfile '%s'"), pack_name);
 313                 output_fd = -1;
 314                 nothread_data.pack_fd = input_fd;
 315         }
 316         the_hash_algo->init_fn(&input_ctx);
 317         return pack_name;
 318 }
 319
 320 static void parse_pack_header(void)
 321 {
 322         struct pack_header *hdr = fill(sizeof(struct pack_header));
 323
 324         /* Header consistency check */
 325         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 326                 die(_("pack signature mismatch"));
 327         if (!pack_version_ok(hdr->hdr_version))
 328                 die(_("pack version %"PRIu32" unsupported"),
 329                         ntohl(hdr->hdr_version));
 330
 331         nr_objects = ntohl(hdr->hdr_entries);
 332         use(sizeof(struct pack_header));
 333 }
 334
 335 static NORETURN void bad_object(off_t offset, const char *format,
 336                        ...) __attribute__((format (printf, 2, 3)));
 337
 338 static NORETURN void bad_object(off_t offset, const char *format, ...)
 339 {
 340         va_list params;
 341         char buf[1024];
 342
 343         va_start(params, format);
 344         vsnprintf(buf, sizeof(buf), format, params);
 345         va_end(params);
 346         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 347             (uintmax_t)offset, buf);
 348 }
 349
 350 static inline struct thread_local *get_thread_data(void)
 351 {
 352         if (HAVE_THREADS) {
 353                 if (threads_active)
 354                         return pthread_getspecific(key);
 355                 assert(!threads_active &&
 356                        "This should only be reached when all threads are gone");
 357         }
 358         return &nothread_data;
 359 }
 360
 361 static void set_thread_data(struct thread_local *data)
 362 {
 363         if (threads_active)
 364                 pthread_setspecific(key, data);
 365 }
 366
 367 static struct base_data *alloc_base_data(void)
 368 {
 369         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 370         base->ref_last = -1;
 371         base->ofs_last = -1;
 372         return base;
 373 }
 374
 375 static void free_base_data(struct base_data *c)
 376 {
 377         if (c->data) {
 378                 FREE_AND_NULL(c->data);
 379                 get_thread_data()->base_cache_used -= c->size;
 380         }
 381 }
 382
 383 static void prune_base_data(struct base_data *retain)
 384 {
 385         struct base_data *b;
 386         struct thread_local *data = get_thread_data();
 387         for (b = data->base_cache;
 388              data->base_cache_used > delta_base_cache_limit && b;
 389              b = b->child) {
 390                 if (b->data && b != retain)
 391                         free_base_data(b);
 392         }
 393 }
 394
 395 static void link_base_data(struct base_data *base, struct base_data *c)
 396 {
 397         if (base)
 398                 base->child = c;
 399         else
 400                 get_thread_data()->base_cache = c;
 401
 402         c->base = base;
 403         c->child = NULL;
 404         if (c->data)
 405                 get_thread_data()->base_cache_used += c->size;
 406         prune_base_data(c);
 407 }
 408
 409 static void unlink_base_data(struct base_data *c)
 410 {
 411         struct base_data *base = c->base;
 412         if (base)
 413                 base->child = NULL;
 414         else
 415                 get_thread_data()->base_cache = NULL;
 416         free_base_data(c);
 417 }
 418
 419 static int is_delta_type(enum object_type type)
 420 {
 421         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 422 }
 423
 424 static void *unpack_entry_data(off_t offset, unsigned long size,
 425                                enum object_type type, struct object_id *oid)
 426 {
 427         static char fixed_buf[8192];
 428         int status;
 429         git_zstream stream;
 430         void *buf;
 431         git_hash_ctx c;
 432         char hdr[32];
 433         int hdrlen;
 434
 435         if (!is_delta_type(type)) {
 436                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %"PRIuMAX,
 437                                    type_name(type),(uintmax_t)size) + 1;
 438                 the_hash_algo->init_fn(&c);
 439                 the_hash_algo->update_fn(&c, hdr, hdrlen);
 440         } else
 441                 oid = NULL;
 442         if (type == OBJ_BLOB && size > big_file_threshold)
 443                 buf = fixed_buf;
 444         else
 445                 buf = xmallocz(size);
 446
 447         memset(&stream, 0, sizeof(stream));
 448         git_inflate_init(&stream);
 449         stream.next_out = buf;
 450         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 451
 452         do {
 453                 unsigned char *last_out = stream.next_out;
 454                 stream.next_in = fill(1);
 455                 stream.avail_in = input_len;
 456                 status = git_inflate(&stream, 0);
 457                 use(input_len - stream.avail_in);
 458                 if (oid)
 459                         the_hash_algo->update_fn(&c, last_out, stream.next_out - last_out);
 460                 if (buf == fixed_buf) {
 461                         stream.next_out = buf;
 462                         stream.avail_out = sizeof(fixed_buf);
 463                 }
 464         } while (status == Z_OK);
 465         if (stream.total_out != size || status != Z_STREAM_END)
 466                 bad_object(offset, _("inflate returned %d"), status);
 467         git_inflate_end(&stream);
 468         if (oid)
 469                 the_hash_algo->final_fn(oid->hash, &c);
 470         return buf == fixed_buf ? NULL : buf;
 471 }
 472
 473 static void *unpack_raw_entry(struct object_entry *obj,
 474                               off_t *ofs_offset,
 475                               struct object_id *ref_oid,
 476                               struct object_id *oid)
 477 {
 478         unsigned char *p;
 479         unsigned long size, c;
 480         off_t base_offset;
 481         unsigned shift;
 482         void *data;
 483
 484         obj->idx.offset = consumed_bytes;
 485         input_crc32 = crc32(0, NULL, 0);
 486
 487         p = fill(1);
 488         c = *p;
 489         use(1);
 490         obj->type = (c >> 4) & 7;
 491         size = (c & 15);
 492         shift = 4;
 493         while (c & 0x80) {
 494                 p = fill(1);
 495                 c = *p;
 496                 use(1);
 497                 size += (c & 0x7f) << shift;
 498                 shift += 7;
 499         }
 500         obj->size = size;
 501
 502         switch (obj->type) {
 503         case OBJ_REF_DELTA:
 504                 hashcpy(ref_oid->hash, fill(the_hash_algo->rawsz));
 505                 use(the_hash_algo->rawsz);
 506                 break;
 507         case OBJ_OFS_DELTA:
 508                 p = fill(1);
 509                 c = *p;
 510                 use(1);
 511                 base_offset = c & 127;
 512                 while (c & 128) {
 513                         base_offset += 1;
 514                         if (!base_offset || MSB(base_offset, 7))
 515                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 516                         p = fill(1);
 517                         c = *p;
 518                         use(1);
 519                         base_offset = (base_offset << 7) + (c & 127);
 520                 }
 521                 *ofs_offset = obj->idx.offset - base_offset;
 522                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 523                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 524                 break;
 525         case OBJ_COMMIT:
 526         case OBJ_TREE:
 527         case OBJ_BLOB:
 528         case OBJ_TAG:
 529                 break;
 530         default:
 531                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 532         }
 533         obj->hdr_size = consumed_bytes - obj->idx.offset;
 534
 535         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, oid);
 536         obj->idx.crc32 = input_crc32;
 537         return data;
 538 }
 539
 540 static void *unpack_data(struct object_entry *obj,
 541                          int (*consume)(const unsigned char *, unsigned long, void *),
 542                          void *cb_data)
 543 {
 544         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 545         off_t len = obj[1].idx.offset - from;
 546         unsigned char *data, *inbuf;
 547         git_zstream stream;
 548         int status;
 549
 550         data = xmallocz(consume ? 64*1024 : obj->size);
 551         inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
 552
 553         memset(&stream, 0, sizeof(stream));
 554         git_inflate_init(&stream);
 555         stream.next_out = data;
 556         stream.avail_out = consume ? 64*1024 : obj->size;
 557
 558         do {
 559                 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
 560                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 561                 if (n < 0)
 562                         die_errno(_("cannot pread pack file"));
 563                 if (!n)
 564                         die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
 565                                "premature end of pack file, %"PRIuMAX" bytes missing",
 566                                (unsigned int)len),
 567                             (uintmax_t)len);
 568                 from += n;
 569                 len -= n;
 570                 stream.next_in = inbuf;
 571                 stream.avail_in = n;
 572                 if (!consume)
 573                         status = git_inflate(&stream, 0);
 574                 else {
 575                         do {
 576                                 status = git_inflate(&stream, 0);
 577                                 if (consume(data, stream.next_out - data, cb_data)) {
 578                                         free(inbuf);
 579                                         free(data);
 580                                         return NULL;
 581                                 }
 582                                 stream.next_out = data;
 583                                 stream.avail_out = 64*1024;
 584                         } while (status == Z_OK && stream.avail_in);
 585                 }
 586         } while (len && status == Z_OK && !stream.avail_in);
 587
 588         /* This has been inflated OK when first encountered, so... */
 589         if (status != Z_STREAM_END || stream.total_out != obj->size)
 590                 die(_("serious inflate inconsistency"));
 591
 592         git_inflate_end(&stream);
 593         free(inbuf);
 594         if (consume) {
 595                 FREE_AND_NULL(data);
 596         }
 597         return data;
 598 }
 599
 600 static void *get_data_from_pack(struct object_entry *obj)
 601 {
 602         return unpack_data(obj, NULL, NULL);
 603 }
 604
 605 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 606                                    enum object_type type1,
 607                                    enum object_type type2)
 608 {
 609         int cmp = type1 - type2;
 610         if (cmp)
 611                 return cmp;
 612         return offset1 < offset2 ? -1 :
 613                offset1 > offset2 ?  1 :
 614                0;
 615 }
 616
 617 static int find_ofs_delta(const off_t offset, enum object_type type)
 618 {
 619         int first = 0, last = nr_ofs_deltas;
 620
 621         while (first < last) {
 622                 int next = first + (last - first) / 2;
 623                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 624                 int cmp;
 625
 626                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 627                                               type, objects[delta->obj_no].type);
 628                 if (!cmp)
 629                         return next;
 630                 if (cmp < 0) {
 631                         last = next;
 632                         continue;
 633                 }
 634                 first = next+1;
 635         }
 636         return -first-1;
 637 }
 638
 639 static void find_ofs_delta_children(off_t offset,
 640                                     int *first_index, int *last_index,
 641                                     enum object_type type)
 642 {
 643         int first = find_ofs_delta(offset, type);
 644         int last = first;
 645         int end = nr_ofs_deltas - 1;
 646
 647         if (first < 0) {
 648                 *first_index = 0;
 649                 *last_index = -1;
 650                 return;
 651         }
 652         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 653                 --first;
 654         while (last < end && ofs_deltas[last + 1].offset == offset)
 655                 ++last;
 656         *first_index = first;
 657         *last_index = last;
 658 }
 659
 660 static int compare_ref_delta_bases(const struct object_id *oid1,
 661                                    const struct object_id *oid2,
 662                                    enum object_type type1,
 663                                    enum object_type type2)
 664 {
 665         int cmp = type1 - type2;
 666         if (cmp)
 667                 return cmp;
 668         return oidcmp(oid1, oid2);
 669 }
 670
 671 static int find_ref_delta(const struct object_id *oid, enum object_type type)
 672 {
 673         int first = 0, last = nr_ref_deltas;
 674
 675         while (first < last) {
 676                 int next = first + (last - first) / 2;
 677                 struct ref_delta_entry *delta = &ref_deltas[next];
 678                 int cmp;
 679
 680                 cmp = compare_ref_delta_bases(oid, &delta->oid,
 681                                               type, objects[delta->obj_no].type);
 682                 if (!cmp)
 683                         return next;
 684                 if (cmp < 0) {
 685                         last = next;
 686                         continue;
 687                 }
 688                 first = next+1;
 689         }
 690         return -first-1;
 691 }
 692
 693 static void find_ref_delta_children(const struct object_id *oid,
 694                                     int *first_index, int *last_index,
 695                                     enum object_type type)
 696 {
 697         int first = find_ref_delta(oid, type);
 698         int last = first;
 699         int end = nr_ref_deltas - 1;
 700
 701         if (first < 0) {
 702                 *first_index = 0;
 703                 *last_index = -1;
 704                 return;
 705         }
 706         while (first > 0 && oideq(&ref_deltas[first - 1].oid, oid))
 707                 --first;
 708         while (last < end && oideq(&ref_deltas[last + 1].oid, oid))
 709                 ++last;
 710         *first_index = first;
 711         *last_index = last;
 712 }
 713
 714 struct compare_data {
 715         struct object_entry *entry;
 716         struct git_istream *st;
 717         unsigned char *buf;
 718         unsigned long buf_size;
 719 };
 720
 721 static int compare_objects(const unsigned char *buf, unsigned long size,
 722                            void *cb_data)
 723 {
 724         struct compare_data *data = cb_data;
 725
 726         if (data->buf_size < size) {
 727                 free(data->buf);
 728                 data->buf = xmalloc(size);
 729                 data->buf_size = size;
 730         }
 731
 732         while (size) {
 733                 ssize_t len = read_istream(data->st, data->buf, size);
 734                 if (len == 0)
 735                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 736                             oid_to_hex(&data->entry->idx.oid));
 737                 if (len < 0)
 738                         die(_("unable to read %s"),
 739                             oid_to_hex(&data->entry->idx.oid));
 740                 if (memcmp(buf, data->buf, len))
 741                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 742                             oid_to_hex(&data->entry->idx.oid));
 743                 size -= len;
 744                 buf += len;
 745         }
 746         return 0;
 747 }
 748
 749 static int check_collison(struct object_entry *entry)
 750 {
 751         struct compare_data data;
 752         enum object_type type;
 753         unsigned long size;
 754
 755         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 756                 return -1;
 757
 758         memset(&data, 0, sizeof(data));
 759         data.entry = entry;
 760         data.st = open_istream(the_repository, &entry->idx.oid, &type, &size,
 761                                NULL);
 762         if (!data.st)
 763                 return -1;
 764         if (size != entry->size || type != entry->type)
 765                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 766                     oid_to_hex(&entry->idx.oid));
 767         unpack_data(entry, compare_objects, &data);
 768         close_istream(data.st);
 769         free(data.buf);
 770         return 0;
 771 }
 772
 773 static void sha1_object(const void *data, struct object_entry *obj_entry,
 774                         unsigned long size, enum object_type type,
 775                         const struct object_id *oid)
 776 {
 777         void *new_data = NULL;
 778         int collision_test_needed = 0;
 779
 780         assert(data || obj_entry);
 781
 782         if (startup_info->have_repository) {
 783                 read_lock();
 784                 collision_test_needed =
 785                         has_object_file_with_flags(oid, OBJECT_INFO_QUICK);
 786                 read_unlock();
 787         }
 788
 789         if (collision_test_needed && !data) {
 790                 read_lock();
 791                 if (!check_collison(obj_entry))
 792                         collision_test_needed = 0;
 793                 read_unlock();
 794         }
 795         if (collision_test_needed) {
 796                 void *has_data;
 797                 enum object_type has_type;
 798                 unsigned long has_size;
 799                 read_lock();
 800                 has_type = oid_object_info(the_repository, oid, &has_size);
 801                 if (has_type < 0)
 802                         die(_("cannot read existing object info %s"), oid_to_hex(oid));
 803                 if (has_type != type || has_size != size)
 804                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 805                 has_data = read_object_file(oid, &has_type, &has_size);
 806                 read_unlock();
 807                 if (!data)
 808                         data = new_data = get_data_from_pack(obj_entry);
 809                 if (!has_data)
 810                         die(_("cannot read existing object %s"), oid_to_hex(oid));
 811                 if (size != has_size || type != has_type ||
 812                     memcmp(data, has_data, size) != 0)
 813                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 814                 free(has_data);
 815         }
 816
 817         if (strict || do_fsck_object) {
 818                 read_lock();
 819                 if (type == OBJ_BLOB) {
 820                         struct blob *blob = lookup_blob(the_repository, oid);
 821                         if (blob)
 822                                 blob->object.flags |= FLAG_CHECKED;
 823                         else
 824                                 die(_("invalid blob object %s"), oid_to_hex(oid));
 825                         if (do_fsck_object &&
 826                             fsck_object(&blob->object, (void *)data, size, &fsck_options))
 827                                 die(_("fsck error in packed object"));
 828                 } else {
 829                         struct object *obj;
 830                         int eaten;
 831                         void *buf = (void *) data;
 832
 833                         assert(data && "data can only be NULL for large _blobs_");
 834
 835                         /*
 836                          * we do not need to free the memory here, as the
 837                          * buf is deleted by the caller.
 838                          */
 839                         obj = parse_object_buffer(the_repository, oid, type,
 840                                                   size, buf,
 841                                                   &eaten);
 842                         if (!obj)
 843                                 die(_("invalid %s"), type_name(type));
 844                         if (do_fsck_object &&
 845                             fsck_object(obj, buf, size, &fsck_options))
 846                                 die(_("fsck error in packed object"));
 847                         if (strict && fsck_walk(obj, NULL, &fsck_options))
 848                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 849
 850                         if (obj->type == OBJ_TREE) {
 851                                 struct tree *item = (struct tree *) obj;
 852                                 item->buffer = NULL;
 853                                 obj->parsed = 0;
 854                         }
 855                         if (obj->type == OBJ_COMMIT) {
 856                                 struct commit *commit = (struct commit *) obj;
 857                                 if (detach_commit_buffer(commit, NULL) != data)
 858                                         BUG("parse_object_buffer transmogrified our buffer");
 859                         }
 860                         obj->flags |= FLAG_CHECKED;
 861                 }
 862                 read_unlock();
 863         }
 864
 865         free(new_data);
 866 }
 867
 868 /*
 869  * This function is part of find_unresolved_deltas(). There are two
 870  * walkers going in the opposite ways.
 871  *
 872  * The first one in find_unresolved_deltas() traverses down from
 873  * parent node to children, deflating nodes along the way. However,
 874  * memory for deflated nodes is limited by delta_base_cache_limit, so
 875  * at some point parent node's deflated content may be freed.
 876  *
 877  * The second walker is this function, which goes from current node up
 878  * to top parent if necessary to deflate the node. In normal
 879  * situation, its parent node would be already deflated, so it just
 880  * needs to apply delta.
 881  *
 882  * In the worst case scenario, parent node is no longer deflated because
 883  * we're running out of delta_base_cache_limit; we need to re-deflate
 884  * parents, possibly up to the top base.
 885  *
 886  * All deflated objects here are subject to be freed if we exceed
 887  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 888  * just need to make sure the last node is not freed.
 889  */
 890 static void *get_base_data(struct base_data *c)
 891 {
 892         if (!c->data) {
 893                 struct object_entry *obj = c->obj;
 894                 struct base_data **delta = NULL;
 895                 int delta_nr = 0, delta_alloc = 0;
 896
 897                 while (is_delta_type(c->obj->type) && !c->data) {
 898                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 899                         delta[delta_nr++] = c;
 900                         c = c->base;
 901                 }
 902                 if (!delta_nr) {
 903                         c->data = get_data_from_pack(obj);
 904                         c->size = obj->size;
 905                         get_thread_data()->base_cache_used += c->size;
 906                         prune_base_data(c);
 907                 }
 908                 for (; delta_nr > 0; delta_nr--) {
 909                         void *base, *raw;
 910                         c = delta[delta_nr - 1];
 911                         obj = c->obj;
 912                         base = get_base_data(c->base);
 913                         raw = get_data_from_pack(obj);
 914                         c->data = patch_delta(
 915                                 base, c->base->size,
 916                                 raw, obj->size,
 917                                 &c->size);
 918                         free(raw);
 919                         if (!c->data)
 920                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 921                         get_thread_data()->base_cache_used += c->size;
 922                         prune_base_data(c);
 923                 }
 924                 free(delta);
 925         }
 926         return c->data;
 927 }
 928
 929 static void resolve_delta(struct object_entry *delta_obj,
 930                           struct base_data *base, struct base_data *result)
 931 {
 932         void *base_data, *delta_data;
 933
 934         if (show_stat) {
 935                 int i = delta_obj - objects;
 936                 int j = base->obj - objects;
 937                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 938                 deepest_delta_lock();
 939                 if (deepest_delta < obj_stat[i].delta_depth)
 940                         deepest_delta = obj_stat[i].delta_depth;
 941                 deepest_delta_unlock();
 942                 obj_stat[i].base_object_no = j;
 943         }
 944         delta_data = get_data_from_pack(delta_obj);
 945         base_data = get_base_data(base);
 946         result->obj = delta_obj;
 947         result->data = patch_delta(base_data, base->size,
 948                                    delta_data, delta_obj->size, &result->size);
 949         free(delta_data);
 950         if (!result->data)
 951                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 952         hash_object_file(the_hash_algo, result->data, result->size,
 953                          type_name(delta_obj->real_type), &delta_obj->idx.oid);
 954         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 955                     &delta_obj->idx.oid);
 956         counter_lock();
 957         nr_resolved_deltas++;
 958         counter_unlock();
 959 }
 960
 961 /*
 962  * Standard boolean compare-and-swap: atomically check whether "*type" is
 963  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 964  * and return false.
 965  */
 966 static int compare_and_swap_type(signed char *type,
 967                                  enum object_type want,
 968                                  enum object_type set)
 969 {
 970         enum object_type old;
 971
 972         type_cas_lock();
 973         old = *type;
 974         if (old == want)
 975                 *type = set;
 976         type_cas_unlock();
 977
 978         return old == want;
 979 }
 980
 981 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 982                                                   struct base_data *prev_base)
 983 {
 984         if (base->ref_last == -1 && base->ofs_last == -1) {
 985                 find_ref_delta_children(&base->obj->idx.oid,
 986                                         &base->ref_first, &base->ref_last,
 987                                         OBJ_REF_DELTA);
 988
 989                 find_ofs_delta_children(base->obj->idx.offset,
 990                                         &base->ofs_first, &base->ofs_last,
 991                                         OBJ_OFS_DELTA);
 992
 993                 if (base->ref_last == -1 && base->ofs_last == -1) {
 994                         free(base->data);
 995                         return NULL;
 996                 }
 997
 998                 link_base_data(prev_base, base);
 999         }
1000
1001         if (base->ref_first <= base->ref_last) {
1002                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1003                 struct base_data *result = alloc_base_data();
1004
1005                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1006                                            base->obj->real_type))
1007                         die("REF_DELTA at offset %"PRIuMAX" already resolved (duplicate base %s?)",
1008                             (uintmax_t)child->idx.offset,
1009                             oid_to_hex(&base->obj->idx.oid));
1010
1011                 resolve_delta(child, base, result);
1012                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1013                         free_base_data(base);
1014
1015                 base->ref_first++;
1016                 return result;
1017         }
1018
1019         if (base->ofs_first <= base->ofs_last) {
1020                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1021                 struct base_data *result = alloc_base_data();
1022
1023                 assert(child->real_type == OBJ_OFS_DELTA);
1024                 child->real_type = base->obj->real_type;
1025                 resolve_delta(child, base, result);
1026                 if (base->ofs_first == base->ofs_last)
1027                         free_base_data(base);
1028
1029                 base->ofs_first++;
1030                 return result;
1031         }
1032
1033         unlink_base_data(base);
1034         return NULL;
1035 }
1036
1037 static void find_unresolved_deltas(struct base_data *base)
1038 {
1039         struct base_data *new_base, *prev_base = NULL;
1040         for (;;) {
1041                 new_base = find_unresolved_deltas_1(base, prev_base);
1042
1043                 if (new_base) {
1044                         prev_base = base;
1045                         base = new_base;
1046                 } else {
1047                         free(base);
1048                         base = prev_base;
1049                         if (!base)
1050                                 return;
1051                         prev_base = base->base;
1052                 }
1053         }
1054 }
1055
1056 static int compare_ofs_delta_entry(const void *a, const void *b)
1057 {
1058         const struct ofs_delta_entry *delta_a = a;
1059         const struct ofs_delta_entry *delta_b = b;
1060
1061         return delta_a->offset < delta_b->offset ? -1 :
1062                delta_a->offset > delta_b->offset ?  1 :
1063                0;
1064 }
1065
1066 static int compare_ref_delta_entry(const void *a, const void *b)
1067 {
1068         const struct ref_delta_entry *delta_a = a;
1069         const struct ref_delta_entry *delta_b = b;
1070
1071         return oidcmp(&delta_a->oid, &delta_b->oid);
1072 }
1073
1074 static void resolve_base(struct object_entry *obj)
1075 {
1076         struct base_data *base_obj = alloc_base_data();
1077         base_obj->obj = obj;
1078         base_obj->data = NULL;
1079         find_unresolved_deltas(base_obj);
1080 }
1081
1082 static void *threaded_second_pass(void *data)
1083 {
1084         set_thread_data(data);
1085         for (;;) {
1086                 int i;
1087                 counter_lock();
1088                 display_progress(progress, nr_resolved_deltas);
1089                 counter_unlock();
1090                 work_lock();
1091                 while (nr_dispatched < nr_objects &&
1092                        is_delta_type(objects[nr_dispatched].type))
1093                         nr_dispatched++;
1094                 if (nr_dispatched >= nr_objects) {
1095                         work_unlock();
1096                         break;
1097                 }
1098                 i = nr_dispatched++;
1099                 work_unlock();
1100
1101                 resolve_base(&objects[i]);
1102         }
1103         return NULL;
1104 }
1105
1106 /*
1107  * First pass:
1108  * - find locations of all objects;
1109  * - calculate SHA1 of all non-delta objects;
1110  * - remember base (SHA1 or offset) for all deltas.
1111  */
1112 static void parse_pack_objects(unsigned char *hash)
1113 {
1114         int i, nr_delays = 0;
1115         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1116         struct object_id ref_delta_oid;
1117         struct stat st;
1118
1119         if (verbose)
1120                 progress = start_progress(
1121                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1122                                 nr_objects);
1123         for (i = 0; i < nr_objects; i++) {
1124                 struct object_entry *obj = &objects[i];
1125                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1126                                               &ref_delta_oid,
1127                                               &obj->idx.oid);
1128                 obj->real_type = obj->type;
1129                 if (obj->type == OBJ_OFS_DELTA) {
1130                         nr_ofs_deltas++;
1131                         ofs_delta->obj_no = i;
1132                         ofs_delta++;
1133                 } else if (obj->type == OBJ_REF_DELTA) {
1134                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1135                         oidcpy(&ref_deltas[nr_ref_deltas].oid, &ref_delta_oid);
1136                         ref_deltas[nr_ref_deltas].obj_no = i;
1137                         nr_ref_deltas++;
1138                 } else if (!data) {
1139                         /* large blobs, check later */
1140                         obj->real_type = OBJ_BAD;
1141                         nr_delays++;
1142                 } else
1143                         sha1_object(data, NULL, obj->size, obj->type,
1144                                     &obj->idx.oid);
1145                 free(data);
1146                 display_progress(progress, i+1);
1147         }
1148         objects[i].idx.offset = consumed_bytes;
1149         stop_progress(&progress);
1150
1151         /* Check pack integrity */
1152         flush();
1153         the_hash_algo->final_fn(hash, &input_ctx);
1154         if (!hasheq(fill(the_hash_algo->rawsz), hash))
1155                 die(_("pack is corrupted (SHA1 mismatch)"));
1156         use(the_hash_algo->rawsz);
1157
1158         /* If input_fd is a file, we should have reached its end now. */
1159         if (fstat(input_fd, &st))
1160                 die_errno(_("cannot fstat packfile"));
1161         if (S_ISREG(st.st_mode) &&
1162                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1163                 die(_("pack has junk at the end"));
1164
1165         for (i = 0; i < nr_objects; i++) {
1166                 struct object_entry *obj = &objects[i];
1167                 if (obj->real_type != OBJ_BAD)
1168                         continue;
1169                 obj->real_type = obj->type;
1170                 sha1_object(NULL, obj, obj->size, obj->type,
1171                             &obj->idx.oid);
1172                 nr_delays--;
1173         }
1174         if (nr_delays)
1175                 die(_("confusion beyond insanity in parse_pack_objects()"));
1176 }
1177
1178 /*
1179  * Second pass:
1180  * - for all non-delta objects, look if it is used as a base for
1181  *   deltas;
1182  * - if used as a base, uncompress the object and apply all deltas,
1183  *   recursively checking if the resulting object is used as a base
1184  *   for some more deltas.
1185  */
1186 static void resolve_deltas(void)
1187 {
1188         int i;
1189
1190         if (!nr_ofs_deltas && !nr_ref_deltas)
1191                 return;
1192
1193         /* Sort deltas by base SHA1/offset for fast searching */
1194         QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1195         QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1196
1197         if (verbose || show_resolving_progress)
1198                 progress = start_progress(_("Resolving deltas"),
1199                                           nr_ref_deltas + nr_ofs_deltas);
1200
1201         nr_dispatched = 0;
1202         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1203                 init_thread();
1204                 for (i = 0; i < nr_threads; i++) {
1205                         int ret = pthread_create(&thread_data[i].thread, NULL,
1206                                                  threaded_second_pass, thread_data + i);
1207                         if (ret)
1208                                 die(_("unable to create thread: %s"),
1209                                     strerror(ret));
1210                 }
1211                 for (i = 0; i < nr_threads; i++)
1212                         pthread_join(thread_data[i].thread, NULL);
1213                 cleanup_thread();
1214                 return;
1215         }
1216
1217         for (i = 0; i < nr_objects; i++) {
1218                 struct object_entry *obj = &objects[i];
1219
1220                 if (is_delta_type(obj->type))
1221                         continue;
1222                 resolve_base(obj);
1223                 display_progress(progress, nr_resolved_deltas);
1224         }
1225 }
1226
1227 /*
1228  * Third pass:
1229  * - append objects to convert thin pack to full pack if required
1230  * - write the final pack hash
1231  */
1232 static void fix_unresolved_deltas(struct hashfile *f);
1233 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_hash)
1234 {
1235         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1236                 stop_progress(&progress);
1237                 /* Flush remaining pack final hash. */
1238                 flush();
1239                 return;
1240         }
1241
1242         if (fix_thin_pack) {
1243                 struct hashfile *f;
1244                 unsigned char read_hash[GIT_MAX_RAWSZ], tail_hash[GIT_MAX_RAWSZ];
1245                 struct strbuf msg = STRBUF_INIT;
1246                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1247                 int nr_objects_initial = nr_objects;
1248                 if (nr_unresolved <= 0)
1249                         die(_("confusion beyond insanity"));
1250                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1251                 memset(objects + nr_objects + 1, 0,
1252                        nr_unresolved * sizeof(*objects));
1253                 f = hashfd(output_fd, curr_pack);
1254                 fix_unresolved_deltas(f);
1255                 strbuf_addf(&msg, Q_("completed with %d local object",
1256                                      "completed with %d local objects",
1257                                      nr_objects - nr_objects_initial),
1258                             nr_objects - nr_objects_initial);
1259                 stop_progress_msg(&progress, msg.buf);
1260                 strbuf_release(&msg);
1261                 finalize_hashfile(f, tail_hash, 0);
1262                 hashcpy(read_hash, pack_hash);
1263                 fixup_pack_header_footer(output_fd, pack_hash,
1264                                          curr_pack, nr_objects,
1265                                          read_hash, consumed_bytes-the_hash_algo->rawsz);
1266                 if (!hasheq(read_hash, tail_hash))
1267                         die(_("Unexpected tail checksum for %s "
1268                               "(disk corruption?)"), curr_pack);
1269         }
1270         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1271                 die(Q_("pack has %d unresolved delta",
1272                        "pack has %d unresolved deltas",
1273                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1274                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1275 }
1276
1277 static int write_compressed(struct hashfile *f, void *in, unsigned int size)
1278 {
1279         git_zstream stream;
1280         int status;
1281         unsigned char outbuf[4096];
1282
1283         git_deflate_init(&stream, zlib_compression_level);
1284         stream.next_in = in;
1285         stream.avail_in = size;
1286
1287         do {
1288                 stream.next_out = outbuf;
1289                 stream.avail_out = sizeof(outbuf);
1290                 status = git_deflate(&stream, Z_FINISH);
1291                 hashwrite(f, outbuf, sizeof(outbuf) - stream.avail_out);
1292         } while (status == Z_OK);
1293
1294         if (status != Z_STREAM_END)
1295                 die(_("unable to deflate appended object (%d)"), status);
1296         size = stream.total_out;
1297         git_deflate_end(&stream);
1298         return size;
1299 }
1300
1301 static struct object_entry *append_obj_to_pack(struct hashfile *f,
1302                                const unsigned char *sha1, void *buf,
1303                                unsigned long size, enum object_type type)
1304 {
1305         struct object_entry *obj = &objects[nr_objects++];
1306         unsigned char header[10];
1307         unsigned long s = size;
1308         int n = 0;
1309         unsigned char c = (type << 4) | (s & 15);
1310         s >>= 4;
1311         while (s) {
1312                 header[n++] = c | 0x80;
1313                 c = s & 0x7f;
1314                 s >>= 7;
1315         }
1316         header[n++] = c;
1317         crc32_begin(f);
1318         hashwrite(f, header, n);
1319         obj[0].size = size;
1320         obj[0].hdr_size = n;
1321         obj[0].type = type;
1322         obj[0].real_type = type;
1323         obj[1].idx.offset = obj[0].idx.offset + n;
1324         obj[1].idx.offset += write_compressed(f, buf, size);
1325         obj[0].idx.crc32 = crc32_end(f);
1326         hashflush(f);
1327         hashcpy(obj->idx.oid.hash, sha1);
1328         return obj;
1329 }
1330
1331 static int delta_pos_compare(const void *_a, const void *_b)
1332 {
1333         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1334         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1335         return a->obj_no - b->obj_no;
1336 }
1337
1338 static void fix_unresolved_deltas(struct hashfile *f)
1339 {
1340         struct ref_delta_entry **sorted_by_pos;
1341         int i;
1342
1343         /*
1344          * Since many unresolved deltas may well be themselves base objects
1345          * for more unresolved deltas, we really want to include the
1346          * smallest number of base objects that would cover as much delta
1347          * as possible by picking the
1348          * trunc deltas first, allowing for other deltas to resolve without
1349          * additional base objects.  Since most base objects are to be found
1350          * before deltas depending on them, a good heuristic is to start
1351          * resolving deltas in the same order as their position in the pack.
1352          */
1353         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1354         for (i = 0; i < nr_ref_deltas; i++)
1355                 sorted_by_pos[i] = &ref_deltas[i];
1356         QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
1357
1358         if (has_promisor_remote()) {
1359                 /*
1360                  * Prefetch the delta bases.
1361                  */
1362                 struct oid_array to_fetch = OID_ARRAY_INIT;
1363                 for (i = 0; i < nr_ref_deltas; i++) {
1364                         struct ref_delta_entry *d = sorted_by_pos[i];
1365                         if (!oid_object_info_extended(the_repository, &d->oid,
1366                                                       NULL,
1367                                                       OBJECT_INFO_FOR_PREFETCH))
1368                                 continue;
1369                         oid_array_append(&to_fetch, &d->oid);
1370                 }
1371                 promisor_remote_get_direct(the_repository,
1372                                            to_fetch.oid, to_fetch.nr);
1373                 oid_array_clear(&to_fetch);
1374         }
1375
1376         for (i = 0; i < nr_ref_deltas; i++) {
1377                 struct ref_delta_entry *d = sorted_by_pos[i];
1378                 enum object_type type;
1379                 struct base_data *base_obj = alloc_base_data();
1380
1381                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1382                         continue;
1383                 base_obj->data = read_object_file(&d->oid, &type,
1384                                                   &base_obj->size);
1385                 if (!base_obj->data)
1386                         continue;
1387
1388                 if (check_object_signature(the_repository, &d->oid,
1389                                            base_obj->data, base_obj->size,
1390                                            type_name(type)))
1391                         die(_("local object %s is corrupt"), oid_to_hex(&d->oid));
1392                 base_obj->obj = append_obj_to_pack(f, d->oid.hash,
1393                                         base_obj->data, base_obj->size, type);
1394                 find_unresolved_deltas(base_obj);
1395                 display_progress(progress, nr_resolved_deltas);
1396         }
1397         free(sorted_by_pos);
1398 }
1399
1400 static const char *derive_filename(const char *pack_name, const char *suffix,
1401                                    struct strbuf *buf)
1402 {
1403         size_t len;
1404         if (!strip_suffix(pack_name, ".pack", &len))
1405                 die(_("packfile name '%s' does not end with '.pack'"),
1406                     pack_name);
1407         strbuf_add(buf, pack_name, len);
1408         strbuf_addch(buf, '.');
1409         strbuf_addstr(buf, suffix);
1410         return buf->buf;
1411 }
1412
1413 static void write_special_file(const char *suffix, const char *msg,
1414                                const char *pack_name, const unsigned char *hash,
1415                                const char **report)
1416 {
1417         struct strbuf name_buf = STRBUF_INIT;
1418         const char *filename;
1419         int fd;
1420         int msg_len = strlen(msg);
1421
1422         if (pack_name)
1423                 filename = derive_filename(pack_name, suffix, &name_buf);
1424         else
1425                 filename = odb_pack_name(&name_buf, hash, suffix);
1426
1427         fd = odb_pack_keep(filename);
1428         if (fd < 0) {
1429                 if (errno != EEXIST)
1430                         die_errno(_("cannot write %s file '%s'"),
1431                                   suffix, filename);
1432         } else {
1433                 if (msg_len > 0) {
1434                         write_or_die(fd, msg, msg_len);
1435                         write_or_die(fd, "\n", 1);
1436                 }
1437                 if (close(fd) != 0)
1438                         die_errno(_("cannot close written %s file '%s'"),
1439                                   suffix, filename);
1440                 if (report)
1441                         *report = suffix;
1442         }
1443         strbuf_release(&name_buf);
1444 }
1445
1446 static void final(const char *final_pack_name, const char *curr_pack_name,
1447                   const char *final_index_name, const char *curr_index_name,
1448                   const char *keep_msg, const char *promisor_msg,
1449                   unsigned char *hash)
1450 {
1451         const char *report = "pack";
1452         struct strbuf pack_name = STRBUF_INIT;
1453         struct strbuf index_name = STRBUF_INIT;
1454         int err;
1455
1456         if (!from_stdin) {
1457                 close(input_fd);
1458         } else {
1459                 fsync_or_die(output_fd, curr_pack_name);
1460                 err = close(output_fd);
1461                 if (err)
1462                         die_errno(_("error while closing pack file"));
1463         }
1464
1465         if (keep_msg)
1466                 write_special_file("keep", keep_msg, final_pack_name, hash,
1467                                    &report);
1468         if (promisor_msg)
1469                 write_special_file("promisor", promisor_msg, final_pack_name,
1470                                    hash, NULL);
1471
1472         if (final_pack_name != curr_pack_name) {
1473                 if (!final_pack_name)
1474                         final_pack_name = odb_pack_name(&pack_name, hash, "pack");
1475                 if (finalize_object_file(curr_pack_name, final_pack_name))
1476                         die(_("cannot store pack file"));
1477         } else if (from_stdin)
1478                 chmod(final_pack_name, 0444);
1479
1480         if (final_index_name != curr_index_name) {
1481                 if (!final_index_name)
1482                         final_index_name = odb_pack_name(&index_name, hash, "idx");
1483                 if (finalize_object_file(curr_index_name, final_index_name))
1484                         die(_("cannot store index file"));
1485         } else
1486                 chmod(final_index_name, 0444);
1487
1488         if (do_fsck_object) {
1489                 struct packed_git *p;
1490                 p = add_packed_git(final_index_name, strlen(final_index_name), 0);
1491                 if (p)
1492                         install_packed_git(the_repository, p);
1493         }
1494
1495         if (!from_stdin) {
1496                 printf("%s\n", hash_to_hex(hash));
1497         } else {
1498                 struct strbuf buf = STRBUF_INIT;
1499
1500                 strbuf_addf(&buf, "%s\t%s\n", report, hash_to_hex(hash));
1501                 write_or_die(1, buf.buf, buf.len);
1502                 strbuf_release(&buf);
1503
1504                 /*
1505                  * Let's just mimic git-unpack-objects here and write
1506                  * the last part of the input buffer to stdout.
1507                  */
1508                 while (input_len) {
1509                         err = xwrite(1, input_buffer + input_offset, input_len);
1510                         if (err <= 0)
1511                                 break;
1512                         input_len -= err;
1513                         input_offset += err;
1514                 }
1515         }
1516
1517         strbuf_release(&index_name);
1518         strbuf_release(&pack_name);
1519 }
1520
1521 static int git_index_pack_config(const char *k, const char *v, void *cb)
1522 {
1523         struct pack_idx_option *opts = cb;
1524
1525         if (!strcmp(k, "pack.indexversion")) {
1526                 opts->version = git_config_int(k, v);
1527                 if (opts->version > 2)
1528                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1529                 return 0;
1530         }
1531         if (!strcmp(k, "pack.threads")) {
1532                 nr_threads = git_config_int(k, v);
1533                 if (nr_threads < 0)
1534                         die(_("invalid number of threads specified (%d)"),
1535                             nr_threads);
1536                 if (!HAVE_THREADS && nr_threads != 1) {
1537                         warning(_("no threads support, ignoring %s"), k);
1538                         nr_threads = 1;
1539                 }
1540                 return 0;
1541         }
1542         return git_default_config(k, v, cb);
1543 }
1544
1545 static int cmp_uint32(const void *a_, const void *b_)
1546 {
1547         uint32_t a = *((uint32_t *)a_);
1548         uint32_t b = *((uint32_t *)b_);
1549
1550         return (a < b) ? -1 : (a != b);
1551 }
1552
1553 static void read_v2_anomalous_offsets(struct packed_git *p,
1554                                       struct pack_idx_option *opts)
1555 {
1556         const uint32_t *idx1, *idx2;
1557         uint32_t i;
1558         const uint32_t hashwords = the_hash_algo->rawsz / sizeof(uint32_t);
1559
1560         /* The address of the 4-byte offset table */
1561         idx1 = (((const uint32_t *)p->index_data)
1562                 + 2 /* 8-byte header */
1563                 + 256 /* fan out */
1564                 + hashwords * p->num_objects /* object ID table */
1565                 + p->num_objects /* CRC32 table */
1566                 );
1567
1568         /* The address of the 8-byte offset table */
1569         idx2 = idx1 + p->num_objects;
1570
1571         for (i = 0; i < p->num_objects; i++) {
1572                 uint32_t off = ntohl(idx1[i]);
1573                 if (!(off & 0x80000000))
1574                         continue;
1575                 off = off & 0x7fffffff;
1576                 check_pack_index_ptr(p, &idx2[off * 2]);
1577                 if (idx2[off * 2])
1578                         continue;
1579                 /*
1580                  * The real offset is ntohl(idx2[off * 2]) in high 4
1581                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1582                  * octets.  But idx2[off * 2] is Zero!!!
1583                  */
1584                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1585                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1586         }
1587
1588         QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32);
1589 }
1590
1591 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1592 {
1593         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1594
1595         if (!p)
1596                 die(_("Cannot open existing pack file '%s'"), pack_name);
1597         if (open_pack_index(p))
1598                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1599
1600         /* Read the attributes from the existing idx file */
1601         opts->version = p->index_version;
1602
1603         if (opts->version == 2)
1604                 read_v2_anomalous_offsets(p, opts);
1605
1606         /*
1607          * Get rid of the idx file as we do not need it anymore.
1608          * NEEDSWORK: extract this bit from free_pack_by_name() in
1609          * sha1-file.c, perhaps?  It shouldn't matter very much as we
1610          * know we haven't installed this pack (hence we never have
1611          * read anything from it).
1612          */
1613         close_pack_index(p);
1614         free(p);
1615 }
1616
1617 static void show_pack_info(int stat_only)
1618 {
1619         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1620         unsigned long *chain_histogram = NULL;
1621
1622         if (deepest_delta)
1623                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1624
1625         for (i = 0; i < nr_objects; i++) {
1626                 struct object_entry *obj = &objects[i];
1627
1628                 if (is_delta_type(obj->type))
1629                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1630                 if (stat_only)
1631                         continue;
1632                 printf("%s %-6s %"PRIuMAX" %"PRIuMAX" %"PRIuMAX,
1633                        oid_to_hex(&obj->idx.oid),
1634                        type_name(obj->real_type), (uintmax_t)obj->size,
1635                        (uintmax_t)(obj[1].idx.offset - obj->idx.offset),
1636                        (uintmax_t)obj->idx.offset);
1637                 if (is_delta_type(obj->type)) {
1638                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1639                         printf(" %u %s", obj_stat[i].delta_depth,
1640                                oid_to_hex(&bobj->idx.oid));
1641                 }
1642                 putchar('\n');
1643         }
1644
1645         if (baseobjects)
1646                 printf_ln(Q_("non delta: %d object",
1647                              "non delta: %d objects",
1648                              baseobjects),
1649                           baseobjects);
1650         for (i = 0; i < deepest_delta; i++) {
1651                 if (!chain_histogram[i])
1652                         continue;
1653                 printf_ln(Q_("chain length = %d: %lu object",
1654                              "chain length = %d: %lu objects",
1655                              chain_histogram[i]),
1656                           i + 1,
1657                           chain_histogram[i]);
1658         }
1659 }
1660
1661 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1662 {
1663         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1664         const char *curr_index;
1665         const char *index_name = NULL, *pack_name = NULL;
1666         const char *keep_msg = NULL;
1667         const char *promisor_msg = NULL;
1668         struct strbuf index_name_buf = STRBUF_INIT;
1669         struct pack_idx_entry **idx_objects;
1670         struct pack_idx_option opts;
1671         unsigned char pack_hash[GIT_MAX_RAWSZ];
1672         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1673         int report_end_of_input = 0;
1674
1675         /*
1676          * index-pack never needs to fetch missing objects except when
1677          * REF_DELTA bases are missing (which are explicitly handled). It only
1678          * accesses the repo to do hash collision checks and to check which
1679          * REF_DELTA bases need to be fetched.
1680          */
1681         fetch_if_missing = 0;
1682
1683         if (argc == 2 && !strcmp(argv[1], "-h"))
1684                 usage(index_pack_usage);
1685
1686         read_replace_refs = 0;
1687         fsck_options.walk = mark_link;
1688
1689         reset_pack_idx_option(&opts);
1690         git_config(git_index_pack_config, &opts);
1691         if (prefix && chdir(prefix))
1692                 die(_("Cannot come back to cwd"));
1693
1694         for (i = 1; i < argc; i++) {
1695                 const char *arg = argv[i];
1696
1697                 if (*arg == '-') {
1698                         if (!strcmp(arg, "--stdin")) {
1699                                 from_stdin = 1;
1700                         } else if (!strcmp(arg, "--fix-thin")) {
1701                                 fix_thin_pack = 1;
1702                         } else if (skip_to_optional_arg(arg, "--strict", &arg)) {
1703                                 strict = 1;
1704                                 do_fsck_object = 1;
1705                                 fsck_set_msg_types(&fsck_options, arg);
1706                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1707                                 strict = 1;
1708                                 check_self_contained_and_connected = 1;
1709                         } else if (!strcmp(arg, "--fsck-objects")) {
1710                                 do_fsck_object = 1;
1711                         } else if (!strcmp(arg, "--verify")) {
1712                                 verify = 1;
1713                         } else if (!strcmp(arg, "--verify-stat")) {
1714                                 verify = 1;
1715                                 show_stat = 1;
1716                         } else if (!strcmp(arg, "--verify-stat-only")) {
1717                                 verify = 1;
1718                                 show_stat = 1;
1719                                 stat_only = 1;
1720                         } else if (skip_to_optional_arg(arg, "--keep", &keep_msg)) {
1721                                 ; /* nothing to do */
1722                         } else if (skip_to_optional_arg(arg, "--promisor", &promisor_msg)) {
1723                                 ; /* already parsed */
1724                         } else if (starts_with(arg, "--threads=")) {
1725                                 char *end;
1726                                 nr_threads = strtoul(arg+10, &end, 0);
1727                                 if (!arg[10] || *end || nr_threads < 0)
1728                                         usage(index_pack_usage);
1729                                 if (!HAVE_THREADS && nr_threads != 1) {
1730                                         warning(_("no threads support, ignoring %s"), arg);
1731                                         nr_threads = 1;
1732                                 }
1733                         } else if (starts_with(arg, "--pack_header=")) {
1734                                 struct pack_header *hdr;
1735                                 char *c;
1736
1737                                 hdr = (struct pack_header *)input_buffer;
1738                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1739                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1740                                 if (*c != ',')
1741                                         die(_("bad %s"), arg);
1742                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1743                                 if (*c)
1744                                         die(_("bad %s"), arg);
1745                                 input_len = sizeof(*hdr);
1746                         } else if (!strcmp(arg, "-v")) {
1747                                 verbose = 1;
1748                         } else if (!strcmp(arg, "--show-resolving-progress")) {
1749                                 show_resolving_progress = 1;
1750                         } else if (!strcmp(arg, "--report-end-of-input")) {
1751                                 report_end_of_input = 1;
1752                         } else if (!strcmp(arg, "-o")) {
1753                                 if (index_name || (i+1) >= argc)
1754                                         usage(index_pack_usage);
1755                                 index_name = argv[++i];
1756                         } else if (starts_with(arg, "--index-version=")) {
1757                                 char *c;
1758                                 opts.version = strtoul(arg + 16, &c, 10);
1759                                 if (opts.version > 2)
1760                                         die(_("bad %s"), arg);
1761                                 if (*c == ',')
1762                                         opts.off32_limit = strtoul(c+1, &c, 0);
1763                                 if (*c || opts.off32_limit & 0x80000000)
1764                                         die(_("bad %s"), arg);
1765                         } else if (skip_prefix(arg, "--max-input-size=", &arg)) {
1766                                 max_input_size = strtoumax(arg, NULL, 10);
1767                         } else
1768                                 usage(index_pack_usage);
1769                         continue;
1770                 }
1771
1772                 if (pack_name)
1773                         usage(index_pack_usage);
1774                 pack_name = arg;
1775         }
1776
1777         if (!pack_name && !from_stdin)
1778                 usage(index_pack_usage);
1779         if (fix_thin_pack && !from_stdin)
1780                 die(_("--fix-thin cannot be used without --stdin"));
1781         if (from_stdin && !startup_info->have_repository)
1782                 die(_("--stdin requires a git repository"));
1783         if (!index_name && pack_name)
1784                 index_name = derive_filename(pack_name, "idx", &index_name_buf);
1785
1786         if (verify) {
1787                 if (!index_name)
1788                         die(_("--verify with no packfile name given"));
1789                 read_idx_option(&opts, index_name);
1790                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1791         }
1792         if (strict)
1793                 opts.flags |= WRITE_IDX_STRICT;
1794
1795         if (HAVE_THREADS && !nr_threads) {
1796                 nr_threads = online_cpus();
1797                 /* An experiment showed that more threads does not mean faster */
1798                 if (nr_threads > 3)
1799                         nr_threads = 3;
1800         }
1801
1802         curr_pack = open_pack_file(pack_name);
1803         parse_pack_header();
1804         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1805         if (show_stat)
1806                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1807         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1808         parse_pack_objects(pack_hash);
1809         if (report_end_of_input)
1810                 write_in_full(2, "\0", 1);
1811         resolve_deltas();
1812         conclude_pack(fix_thin_pack, curr_pack, pack_hash);
1813         free(ofs_deltas);
1814         free(ref_deltas);
1815         if (strict)
1816                 foreign_nr = check_objects();
1817
1818         if (show_stat)
1819                 show_pack_info(stat_only);
1820
1821         ALLOC_ARRAY(idx_objects, nr_objects);
1822         for (i = 0; i < nr_objects; i++)
1823                 idx_objects[i] = &objects[i].idx;
1824         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_hash);
1825         free(idx_objects);
1826
1827         if (!verify)
1828                 final(pack_name, curr_pack,
1829                       index_name, curr_index,
1830                       keep_msg, promisor_msg,
1831                       pack_hash);
1832         else
1833                 close(input_fd);
1834
1835         if (do_fsck_object && fsck_finish(&fsck_options))
1836                 die(_("fsck error in pack objects"));
1837
1838         free(objects);
1839         strbuf_release(&index_name_buf);
1840         if (pack_name == NULL)
1841                 free((void *) curr_pack);
1842         if (index_name == NULL)
1843                 free((void *) curr_index);
1844
1845         /*
1846          * Let the caller know this pack is not self contained
1847          */
1848         if (check_self_contained_and_connected && foreign_nr)
1849                 return 1;
1850
1851         return 0;
1852 }