git.oblomov.eu Git - git/blob - 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
  18 static const char index_pack_usage[] =
  19 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  20
  21 struct object_entry {
  22         struct pack_idx_entry idx;
  23         unsigned long size;
  24         unsigned char hdr_size;
  25         signed char type;
  26         signed char real_type;
  27 };
  28
  29 struct object_stat {
  30         unsigned delta_depth;
  31         int base_object_no;
  32 };
  33
  34 struct base_data {
  35         struct base_data *base;
  36         struct base_data *child;
  37         struct object_entry *obj;
  38         void *data;
  39         unsigned long size;
  40         int ref_first, ref_last;
  41         int ofs_first, ofs_last;
  42 };
  43
  44 struct thread_local {
  45 #ifndef NO_PTHREADS
  46         pthread_t thread;
  47 #endif
  48         struct base_data *base_cache;
  49         size_t base_cache_used;
  50         int pack_fd;
  51 };
  52
  53 /* Remember to update object flag allocation in object.h */
  54 #define FLAG_LINK (1u<<20)
  55 #define FLAG_CHECKED (1u<<21)
  56
  57 struct ofs_delta_entry {
  58         off_t offset;
  59         int obj_no;
  60 };
  61
  62 struct ref_delta_entry {
  63         struct object_id oid;
  64         int obj_no;
  65 };
  66
  67 static struct object_entry *objects;
  68 static struct object_stat *obj_stat;
  69 static struct ofs_delta_entry *ofs_deltas;
  70 static struct ref_delta_entry *ref_deltas;
  71 static struct thread_local nothread_data;
  72 static int nr_objects;
  73 static int nr_ofs_deltas;
  74 static int nr_ref_deltas;
  75 static int ref_deltas_alloc;
  76 static int nr_resolved_deltas;
  77 static int nr_threads;
  78
  79 static int from_stdin;
  80 static int strict;
  81 static int do_fsck_object;
  82 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  83 static int verbose;
  84 static int show_resolving_progress;
  85 static int show_stat;
  86 static int check_self_contained_and_connected;
  87
  88 static struct progress *progress;
  89
  90 /* We always read in 4kB chunks. */
  91 static unsigned char input_buffer[4096];
  92 static unsigned int input_offset, input_len;
  93 static off_t consumed_bytes;
  94 static off_t max_input_size;
  95 static unsigned deepest_delta;
  96 static git_hash_ctx input_ctx;
  97 static uint32_t input_crc32;
  98 static int input_fd, output_fd;
  99 static const char *curr_pack;
 100
 101 #ifndef NO_PTHREADS
 102
 103 static struct thread_local *thread_data;
 104 static int nr_dispatched;
 105 static int threads_active;
 106
 107 static pthread_mutex_t read_mutex;
 108 #define read_lock()             lock_mutex(&read_mutex)
 109 #define read_unlock()           unlock_mutex(&read_mutex)
 110
 111 static pthread_mutex_t counter_mutex;
 112 #define counter_lock()          lock_mutex(&counter_mutex)
 113 #define counter_unlock()        unlock_mutex(&counter_mutex)
 114
 115 static pthread_mutex_t work_mutex;
 116 #define work_lock()             lock_mutex(&work_mutex)
 117 #define work_unlock()           unlock_mutex(&work_mutex)
 118
 119 static pthread_mutex_t deepest_delta_mutex;
 120 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 121 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 122
 123 static pthread_mutex_t type_cas_mutex;
 124 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 125 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 126
 127 static pthread_key_t key;
 128
 129 static inline void lock_mutex(pthread_mutex_t *mutex)
 130 {
 131         if (threads_active)
 132                 pthread_mutex_lock(mutex);
 133 }
 134
 135 static inline void unlock_mutex(pthread_mutex_t *mutex)
 136 {
 137         if (threads_active)
 138                 pthread_mutex_unlock(mutex);
 139 }
 140
 141 /*
 142  * Mutex and conditional variable can't be statically-initialized on Windows.
 143  */
 144 static void init_thread(void)
 145 {
 146         int i;
 147         init_recursive_mutex(&read_mutex);
 148         pthread_mutex_init(&counter_mutex, NULL);
 149         pthread_mutex_init(&work_mutex, NULL);
 150         pthread_mutex_init(&type_cas_mutex, NULL);
 151         if (show_stat)
 152                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 153         pthread_key_create(&key, NULL);
 154         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 155         for (i = 0; i < nr_threads; i++) {
 156                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 157                 if (thread_data[i].pack_fd == -1)
 158                         die_errno(_("unable to open %s"), curr_pack);
 159         }
 160
 161         threads_active = 1;
 162 }
 163
 164 static void cleanup_thread(void)
 165 {
 166         int i;
 167         if (!threads_active)
 168                 return;
 169         threads_active = 0;
 170         pthread_mutex_destroy(&read_mutex);
 171         pthread_mutex_destroy(&counter_mutex);
 172         pthread_mutex_destroy(&work_mutex);
 173         pthread_mutex_destroy(&type_cas_mutex);
 174         if (show_stat)
 175                 pthread_mutex_destroy(&deepest_delta_mutex);
 176         for (i = 0; i < nr_threads; i++)
 177                 close(thread_data[i].pack_fd);
 178         pthread_key_delete(key);
 179         free(thread_data);
 180 }
 181
 182 #else
 183
 184 #define read_lock()
 185 #define read_unlock()
 186
 187 #define counter_lock()
 188 #define counter_unlock()
 189
 190 #define work_lock()
 191 #define work_unlock()
 192
 193 #define deepest_delta_lock()
 194 #define deepest_delta_unlock()
 195
 196 #define type_cas_lock()
 197 #define type_cas_unlock()
 198
 199 #endif
 200
 201
 202 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 203 {
 204         if (!obj)
 205                 return -1;
 206
 207         if (type != OBJ_ANY && obj->type != type)
 208                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 209
 210         obj->flags |= FLAG_LINK;
 211         return 0;
 212 }
 213
 214 /* The content of each linked object must have been checked
 215    or it must be already present in the object database */
 216 static unsigned check_object(struct object *obj)
 217 {
 218         if (!obj)
 219                 return 0;
 220
 221         if (!(obj->flags & FLAG_LINK))
 222                 return 0;
 223
 224         if (!(obj->flags & FLAG_CHECKED)) {
 225                 unsigned long size;
 226                 int type = oid_object_info(the_repository, &obj->oid, &size);
 227                 if (type <= 0)
 228                         die(_("did not receive expected object %s"),
 229                               oid_to_hex(&obj->oid));
 230                 if (type != obj->type)
 231                         die(_("object %s: expected type %s, found %s"),
 232                             oid_to_hex(&obj->oid),
 233                             type_name(obj->type), type_name(type));
 234                 obj->flags |= FLAG_CHECKED;
 235                 return 1;
 236         }
 237
 238         return 0;
 239 }
 240
 241 static unsigned check_objects(void)
 242 {
 243         unsigned i, max, foreign_nr = 0;
 244
 245         max = get_max_object_index();
 246         for (i = 0; i < max; i++)
 247                 foreign_nr += check_object(get_indexed_object(i));
 248         return foreign_nr;
 249 }
 250
 251
 252 /* Discard current buffer used content. */
 253 static void flush(void)
 254 {
 255         if (input_offset) {
 256                 if (output_fd >= 0)
 257                         write_or_die(output_fd, input_buffer, input_offset);
 258                 the_hash_algo->update_fn(&input_ctx, input_buffer, input_offset);
 259                 memmove(input_buffer, input_buffer + input_offset, input_len);
 260                 input_offset = 0;
 261         }
 262 }
 263
 264 /*
 265  * Make sure at least "min" bytes are available in the buffer, and
 266  * return the pointer to the buffer.
 267  */
 268 static void *fill(int min)
 269 {
 270         if (min <= input_len)
 271                 return input_buffer + input_offset;
 272         if (min > sizeof(input_buffer))
 273                 die(Q_("cannot fill %d byte",
 274                        "cannot fill %d bytes",
 275                        min),
 276                     min);
 277         flush();
 278         do {
 279                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 280                                 sizeof(input_buffer) - input_len);
 281                 if (ret <= 0) {
 282                         if (!ret)
 283                                 die(_("early EOF"));
 284                         die_errno(_("read error on input"));
 285                 }
 286                 input_len += ret;
 287                 if (from_stdin)
 288                         display_throughput(progress, consumed_bytes + input_len);
 289         } while (input_len < min);
 290         return input_buffer;
 291 }
 292
 293 static void use(int bytes)
 294 {
 295         if (bytes > input_len)
 296                 die(_("used more bytes than were available"));
 297         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 298         input_len -= bytes;
 299         input_offset += bytes;
 300
 301         /* make sure off_t is sufficiently large not to wrap */
 302         if (signed_add_overflows(consumed_bytes, bytes))
 303                 die(_("pack too large for current definition of off_t"));
 304         consumed_bytes += bytes;
 305         if (max_input_size && consumed_bytes > max_input_size)
 306                 die(_("pack exceeds maximum allowed size"));
 307 }
 308
 309 static const char *open_pack_file(const char *pack_name)
 310 {
 311         if (from_stdin) {
 312                 input_fd = 0;
 313                 if (!pack_name) {
 314                         struct strbuf tmp_file = STRBUF_INIT;
 315                         output_fd = odb_mkstemp(&tmp_file,
 316                                                 "pack/tmp_pack_XXXXXX");
 317                         pack_name = strbuf_detach(&tmp_file, NULL);
 318                 } else {
 319                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 320                         if (output_fd < 0)
 321                                 die_errno(_("unable to create '%s'"), pack_name);
 322                 }
 323                 nothread_data.pack_fd = output_fd;
 324         } else {
 325                 input_fd = open(pack_name, O_RDONLY);
 326                 if (input_fd < 0)
 327                         die_errno(_("cannot open packfile '%s'"), pack_name);
 328                 output_fd = -1;
 329                 nothread_data.pack_fd = input_fd;
 330         }
 331         the_hash_algo->init_fn(&input_ctx);
 332         return pack_name;
 333 }
 334
 335 static void parse_pack_header(void)
 336 {
 337         struct pack_header *hdr = fill(sizeof(struct pack_header));
 338
 339         /* Header consistency check */
 340         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 341                 die(_("pack signature mismatch"));
 342         if (!pack_version_ok(hdr->hdr_version))
 343                 die(_("pack version %"PRIu32" unsupported"),
 344                         ntohl(hdr->hdr_version));
 345
 346         nr_objects = ntohl(hdr->hdr_entries);
 347         use(sizeof(struct pack_header));
 348 }
 349
 350 static NORETURN void bad_object(off_t offset, const char *format,
 351                        ...) __attribute__((format (printf, 2, 3)));
 352
 353 static NORETURN void bad_object(off_t offset, const char *format, ...)
 354 {
 355         va_list params;
 356         char buf[1024];
 357
 358         va_start(params, format);
 359         vsnprintf(buf, sizeof(buf), format, params);
 360         va_end(params);
 361         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 362             (uintmax_t)offset, buf);
 363 }
 364
 365 static inline struct thread_local *get_thread_data(void)
 366 {
 367 #ifndef NO_PTHREADS
 368         if (threads_active)
 369                 return pthread_getspecific(key);
 370         assert(!threads_active &&
 371                "This should only be reached when all threads are gone");
 372 #endif
 373         return &nothread_data;
 374 }
 375
 376 #ifndef NO_PTHREADS
 377 static void set_thread_data(struct thread_local *data)
 378 {
 379         if (threads_active)
 380                 pthread_setspecific(key, data);
 381 }
 382 #endif
 383
 384 static struct base_data *alloc_base_data(void)
 385 {
 386         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 387         base->ref_last = -1;
 388         base->ofs_last = -1;
 389         return base;
 390 }
 391
 392 static void free_base_data(struct base_data *c)
 393 {
 394         if (c->data) {
 395                 FREE_AND_NULL(c->data);
 396                 get_thread_data()->base_cache_used -= c->size;
 397         }
 398 }
 399
 400 static void prune_base_data(struct base_data *retain)
 401 {
 402         struct base_data *b;
 403         struct thread_local *data = get_thread_data();
 404         for (b = data->base_cache;
 405              data->base_cache_used > delta_base_cache_limit && b;
 406              b = b->child) {
 407                 if (b->data && b != retain)
 408                         free_base_data(b);
 409         }
 410 }
 411
 412 static void link_base_data(struct base_data *base, struct base_data *c)
 413 {
 414         if (base)
 415                 base->child = c;
 416         else
 417                 get_thread_data()->base_cache = c;
 418
 419         c->base = base;
 420         c->child = NULL;
 421         if (c->data)
 422                 get_thread_data()->base_cache_used += c->size;
 423         prune_base_data(c);
 424 }
 425
 426 static void unlink_base_data(struct base_data *c)
 427 {
 428         struct base_data *base = c->base;
 429         if (base)
 430                 base->child = NULL;
 431         else
 432                 get_thread_data()->base_cache = NULL;
 433         free_base_data(c);
 434 }
 435
 436 static int is_delta_type(enum object_type type)
 437 {
 438         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 439 }
 440
 441 static void *unpack_entry_data(off_t offset, unsigned long size,
 442                                enum object_type type, struct object_id *oid)
 443 {
 444         static char fixed_buf[8192];
 445         int status;
 446         git_zstream stream;
 447         void *buf;
 448         git_hash_ctx c;
 449         char hdr[32];
 450         int hdrlen;
 451
 452         if (!is_delta_type(type)) {
 453                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %"PRIuMAX,
 454                                    type_name(type),(uintmax_t)size) + 1;
 455                 the_hash_algo->init_fn(&c);
 456                 the_hash_algo->update_fn(&c, hdr, hdrlen);
 457         } else
 458                 oid = NULL;
 459         if (type == OBJ_BLOB && size > big_file_threshold)
 460                 buf = fixed_buf;
 461         else
 462                 buf = xmallocz(size);
 463
 464         memset(&stream, 0, sizeof(stream));
 465         git_inflate_init(&stream);
 466         stream.next_out = buf;
 467         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 468
 469         do {
 470                 unsigned char *last_out = stream.next_out;
 471                 stream.next_in = fill(1);
 472                 stream.avail_in = input_len;
 473                 status = git_inflate(&stream, 0);
 474                 use(input_len - stream.avail_in);
 475                 if (oid)
 476                         the_hash_algo->update_fn(&c, last_out, stream.next_out - last_out);
 477                 if (buf == fixed_buf) {
 478                         stream.next_out = buf;
 479                         stream.avail_out = sizeof(fixed_buf);
 480                 }
 481         } while (status == Z_OK);
 482         if (stream.total_out != size || status != Z_STREAM_END)
 483                 bad_object(offset, _("inflate returned %d"), status);
 484         git_inflate_end(&stream);
 485         if (oid)
 486                 the_hash_algo->final_fn(oid->hash, &c);
 487         return buf == fixed_buf ? NULL : buf;
 488 }
 489
 490 static void *unpack_raw_entry(struct object_entry *obj,
 491                               off_t *ofs_offset,
 492                               struct object_id *ref_oid,
 493                               struct object_id *oid)
 494 {
 495         unsigned char *p;
 496         unsigned long size, c;
 497         off_t base_offset;
 498         unsigned shift;
 499         void *data;
 500
 501         obj->idx.offset = consumed_bytes;
 502         input_crc32 = crc32(0, NULL, 0);
 503
 504         p = fill(1);
 505         c = *p;
 506         use(1);
 507         obj->type = (c >> 4) & 7;
 508         size = (c & 15);
 509         shift = 4;
 510         while (c & 0x80) {
 511                 p = fill(1);
 512                 c = *p;
 513                 use(1);
 514                 size += (c & 0x7f) << shift;
 515                 shift += 7;
 516         }
 517         obj->size = size;
 518
 519         switch (obj->type) {
 520         case OBJ_REF_DELTA:
 521                 hashcpy(ref_oid->hash, fill(the_hash_algo->rawsz));
 522                 use(the_hash_algo->rawsz);
 523                 break;
 524         case OBJ_OFS_DELTA:
 525                 p = fill(1);
 526                 c = *p;
 527                 use(1);
 528                 base_offset = c & 127;
 529                 while (c & 128) {
 530                         base_offset += 1;
 531                         if (!base_offset || MSB(base_offset, 7))
 532                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 533                         p = fill(1);
 534                         c = *p;
 535                         use(1);
 536                         base_offset = (base_offset << 7) + (c & 127);
 537                 }
 538                 *ofs_offset = obj->idx.offset - base_offset;
 539                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 540                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 541                 break;
 542         case OBJ_COMMIT:
 543         case OBJ_TREE:
 544         case OBJ_BLOB:
 545         case OBJ_TAG:
 546                 break;
 547         default:
 548                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 549         }
 550         obj->hdr_size = consumed_bytes - obj->idx.offset;
 551
 552         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, oid);
 553         obj->idx.crc32 = input_crc32;
 554         return data;
 555 }
 556
 557 static void *unpack_data(struct object_entry *obj,
 558                          int (*consume)(const unsigned char *, unsigned long, void *),
 559                          void *cb_data)
 560 {
 561         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 562         off_t len = obj[1].idx.offset - from;
 563         unsigned char *data, *inbuf;
 564         git_zstream stream;
 565         int status;
 566
 567         data = xmallocz(consume ? 64*1024 : obj->size);
 568         inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
 569
 570         memset(&stream, 0, sizeof(stream));
 571         git_inflate_init(&stream);
 572         stream.next_out = data;
 573         stream.avail_out = consume ? 64*1024 : obj->size;
 574
 575         do {
 576                 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
 577                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 578                 if (n < 0)
 579                         die_errno(_("cannot pread pack file"));
 580                 if (!n)
 581                         die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
 582                                "premature end of pack file, %"PRIuMAX" bytes missing",
 583                                (unsigned int)len),
 584                             (uintmax_t)len);
 585                 from += n;
 586                 len -= n;
 587                 stream.next_in = inbuf;
 588                 stream.avail_in = n;
 589                 if (!consume)
 590                         status = git_inflate(&stream, 0);
 591                 else {
 592                         do {
 593                                 status = git_inflate(&stream, 0);
 594                                 if (consume(data, stream.next_out - data, cb_data)) {
 595                                         free(inbuf);
 596                                         free(data);
 597                                         return NULL;
 598                                 }
 599                                 stream.next_out = data;
 600                                 stream.avail_out = 64*1024;
 601                         } while (status == Z_OK && stream.avail_in);
 602                 }
 603         } while (len && status == Z_OK && !stream.avail_in);
 604
 605         /* This has been inflated OK when first encountered, so... */
 606         if (status != Z_STREAM_END || stream.total_out != obj->size)
 607                 die(_("serious inflate inconsistency"));
 608
 609         git_inflate_end(&stream);
 610         free(inbuf);
 611         if (consume) {
 612                 FREE_AND_NULL(data);
 613         }
 614         return data;
 615 }
 616
 617 static void *get_data_from_pack(struct object_entry *obj)
 618 {
 619         return unpack_data(obj, NULL, NULL);
 620 }
 621
 622 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 623                                    enum object_type type1,
 624                                    enum object_type type2)
 625 {
 626         int cmp = type1 - type2;
 627         if (cmp)
 628                 return cmp;
 629         return offset1 < offset2 ? -1 :
 630                offset1 > offset2 ?  1 :
 631                0;
 632 }
 633
 634 static int find_ofs_delta(const off_t offset, enum object_type type)
 635 {
 636         int first = 0, last = nr_ofs_deltas;
 637
 638         while (first < last) {
 639                 int next = first + (last - first) / 2;
 640                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 641                 int cmp;
 642
 643                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 644                                               type, objects[delta->obj_no].type);
 645                 if (!cmp)
 646                         return next;
 647                 if (cmp < 0) {
 648                         last = next;
 649                         continue;
 650                 }
 651                 first = next+1;
 652         }
 653         return -first-1;
 654 }
 655
 656 static void find_ofs_delta_children(off_t offset,
 657                                     int *first_index, int *last_index,
 658                                     enum object_type type)
 659 {
 660         int first = find_ofs_delta(offset, type);
 661         int last = first;
 662         int end = nr_ofs_deltas - 1;
 663
 664         if (first < 0) {
 665                 *first_index = 0;
 666                 *last_index = -1;
 667                 return;
 668         }
 669         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 670                 --first;
 671         while (last < end && ofs_deltas[last + 1].offset == offset)
 672                 ++last;
 673         *first_index = first;
 674         *last_index = last;
 675 }
 676
 677 static int compare_ref_delta_bases(const struct object_id *oid1,
 678                                    const struct object_id *oid2,
 679                                    enum object_type type1,
 680                                    enum object_type type2)
 681 {
 682         int cmp = type1 - type2;
 683         if (cmp)
 684                 return cmp;
 685         return oidcmp(oid1, oid2);
 686 }
 687
 688 static int find_ref_delta(const struct object_id *oid, enum object_type type)
 689 {
 690         int first = 0, last = nr_ref_deltas;
 691
 692         while (first < last) {
 693                 int next = first + (last - first) / 2;
 694                 struct ref_delta_entry *delta = &ref_deltas[next];
 695                 int cmp;
 696
 697                 cmp = compare_ref_delta_bases(oid, &delta->oid,
 698                                               type, objects[delta->obj_no].type);
 699                 if (!cmp)
 700                         return next;
 701                 if (cmp < 0) {
 702                         last = next;
 703                         continue;
 704                 }
 705                 first = next+1;
 706         }
 707         return -first-1;
 708 }
 709
 710 static void find_ref_delta_children(const struct object_id *oid,
 711                                     int *first_index, int *last_index,
 712                                     enum object_type type)
 713 {
 714         int first = find_ref_delta(oid, type);
 715         int last = first;
 716         int end = nr_ref_deltas - 1;
 717
 718         if (first < 0) {
 719                 *first_index = 0;
 720                 *last_index = -1;
 721                 return;
 722         }
 723         while (first > 0 && oideq(&ref_deltas[first - 1].oid, oid))
 724                 --first;
 725         while (last < end && oideq(&ref_deltas[last + 1].oid, oid))
 726                 ++last;
 727         *first_index = first;
 728         *last_index = last;
 729 }
 730
 731 struct compare_data {
 732         struct object_entry *entry;
 733         struct git_istream *st;
 734         unsigned char *buf;
 735         unsigned long buf_size;
 736 };
 737
 738 static int compare_objects(const unsigned char *buf, unsigned long size,
 739                            void *cb_data)
 740 {
 741         struct compare_data *data = cb_data;
 742
 743         if (data->buf_size < size) {
 744                 free(data->buf);
 745                 data->buf = xmalloc(size);
 746                 data->buf_size = size;
 747         }
 748
 749         while (size) {
 750                 ssize_t len = read_istream(data->st, data->buf, size);
 751                 if (len == 0)
 752                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 753                             oid_to_hex(&data->entry->idx.oid));
 754                 if (len < 0)
 755                         die(_("unable to read %s"),
 756                             oid_to_hex(&data->entry->idx.oid));
 757                 if (memcmp(buf, data->buf, len))
 758                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 759                             oid_to_hex(&data->entry->idx.oid));
 760                 size -= len;
 761                 buf += len;
 762         }
 763         return 0;
 764 }
 765
 766 static int check_collison(struct object_entry *entry)
 767 {
 768         struct compare_data data;
 769         enum object_type type;
 770         unsigned long size;
 771
 772         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 773                 return -1;
 774
 775         memset(&data, 0, sizeof(data));
 776         data.entry = entry;
 777         data.st = open_istream(&entry->idx.oid, &type, &size, NULL);
 778         if (!data.st)
 779                 return -1;
 780         if (size != entry->size || type != entry->type)
 781                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 782                     oid_to_hex(&entry->idx.oid));
 783         unpack_data(entry, compare_objects, &data);
 784         close_istream(data.st);
 785         free(data.buf);
 786         return 0;
 787 }
 788
 789 static void sha1_object(const void *data, struct object_entry *obj_entry,
 790                         unsigned long size, enum object_type type,
 791                         const struct object_id *oid)
 792 {
 793         void *new_data = NULL;
 794         int collision_test_needed = 0;
 795
 796         assert(data || obj_entry);
 797
 798         if (startup_info->have_repository) {
 799                 read_lock();
 800                 collision_test_needed =
 801                         has_sha1_file_with_flags(oid->hash, OBJECT_INFO_QUICK);
 802                 read_unlock();
 803         }
 804
 805         if (collision_test_needed && !data) {
 806                 read_lock();
 807                 if (!check_collison(obj_entry))
 808                         collision_test_needed = 0;
 809                 read_unlock();
 810         }
 811         if (collision_test_needed) {
 812                 void *has_data;
 813                 enum object_type has_type;
 814                 unsigned long has_size;
 815                 read_lock();
 816                 has_type = oid_object_info(the_repository, oid, &has_size);
 817                 if (has_type < 0)
 818                         die(_("cannot read existing object info %s"), oid_to_hex(oid));
 819                 if (has_type != type || has_size != size)
 820                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 821                 has_data = read_object_file(oid, &has_type, &has_size);
 822                 read_unlock();
 823                 if (!data)
 824                         data = new_data = get_data_from_pack(obj_entry);
 825                 if (!has_data)
 826                         die(_("cannot read existing object %s"), oid_to_hex(oid));
 827                 if (size != has_size || type != has_type ||
 828                     memcmp(data, has_data, size) != 0)
 829                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 830                 free(has_data);
 831         }
 832
 833         if (strict || do_fsck_object) {
 834                 read_lock();
 835                 if (type == OBJ_BLOB) {
 836                         struct blob *blob = lookup_blob(the_repository, oid);
 837                         if (blob)
 838                                 blob->object.flags |= FLAG_CHECKED;
 839                         else
 840                                 die(_("invalid blob object %s"), oid_to_hex(oid));
 841                         if (do_fsck_object &&
 842                             fsck_object(&blob->object, (void *)data, size, &fsck_options))
 843                                 die(_("fsck error in packed object"));
 844                 } else {
 845                         struct object *obj;
 846                         int eaten;
 847                         void *buf = (void *) data;
 848
 849                         assert(data && "data can only be NULL for large _blobs_");
 850
 851                         /*
 852                          * we do not need to free the memory here, as the
 853                          * buf is deleted by the caller.
 854                          */
 855                         obj = parse_object_buffer(the_repository, oid, type,
 856                                                   size, buf,
 857                                                   &eaten);
 858                         if (!obj)
 859                                 die(_("invalid %s"), type_name(type));
 860                         if (do_fsck_object &&
 861                             fsck_object(obj, buf, size, &fsck_options))
 862                                 die(_("fsck error in packed object"));
 863                         if (strict && fsck_walk(obj, NULL, &fsck_options))
 864                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 865
 866                         if (obj->type == OBJ_TREE) {
 867                                 struct tree *item = (struct tree *) obj;
 868                                 item->buffer = NULL;
 869                                 obj->parsed = 0;
 870                         }
 871                         if (obj->type == OBJ_COMMIT) {
 872                                 struct commit *commit = (struct commit *) obj;
 873                                 if (detach_commit_buffer(commit, NULL) != data)
 874                                         BUG("parse_object_buffer transmogrified our buffer");
 875                         }
 876                         obj->flags |= FLAG_CHECKED;
 877                 }
 878                 read_unlock();
 879         }
 880
 881         free(new_data);
 882 }
 883
 884 /*
 885  * This function is part of find_unresolved_deltas(). There are two
 886  * walkers going in the opposite ways.
 887  *
 888  * The first one in find_unresolved_deltas() traverses down from
 889  * parent node to children, deflating nodes along the way. However,
 890  * memory for deflated nodes is limited by delta_base_cache_limit, so
 891  * at some point parent node's deflated content may be freed.
 892  *
 893  * The second walker is this function, which goes from current node up
 894  * to top parent if necessary to deflate the node. In normal
 895  * situation, its parent node would be already deflated, so it just
 896  * needs to apply delta.
 897  *
 898  * In the worst case scenario, parent node is no longer deflated because
 899  * we're running out of delta_base_cache_limit; we need to re-deflate
 900  * parents, possibly up to the top base.
 901  *
 902  * All deflated objects here are subject to be freed if we exceed
 903  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 904  * just need to make sure the last node is not freed.
 905  */
 906 static void *get_base_data(struct base_data *c)
 907 {
 908         if (!c->data) {
 909                 struct object_entry *obj = c->obj;
 910                 struct base_data **delta = NULL;
 911                 int delta_nr = 0, delta_alloc = 0;
 912
 913                 while (is_delta_type(c->obj->type) && !c->data) {
 914                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 915                         delta[delta_nr++] = c;
 916                         c = c->base;
 917                 }
 918                 if (!delta_nr) {
 919                         c->data = get_data_from_pack(obj);
 920                         c->size = obj->size;
 921                         get_thread_data()->base_cache_used += c->size;
 922                         prune_base_data(c);
 923                 }
 924                 for (; delta_nr > 0; delta_nr--) {
 925                         void *base, *raw;
 926                         c = delta[delta_nr - 1];
 927                         obj = c->obj;
 928                         base = get_base_data(c->base);
 929                         raw = get_data_from_pack(obj);
 930                         c->data = patch_delta(
 931                                 base, c->base->size,
 932                                 raw, obj->size,
 933                                 &c->size);
 934                         free(raw);
 935                         if (!c->data)
 936                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 937                         get_thread_data()->base_cache_used += c->size;
 938                         prune_base_data(c);
 939                 }
 940                 free(delta);
 941         }
 942         return c->data;
 943 }
 944
 945 static void resolve_delta(struct object_entry *delta_obj,
 946                           struct base_data *base, struct base_data *result)
 947 {
 948         void *base_data, *delta_data;
 949
 950         if (show_stat) {
 951                 int i = delta_obj - objects;
 952                 int j = base->obj - objects;
 953                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 954                 deepest_delta_lock();
 955                 if (deepest_delta < obj_stat[i].delta_depth)
 956                         deepest_delta = obj_stat[i].delta_depth;
 957                 deepest_delta_unlock();
 958                 obj_stat[i].base_object_no = j;
 959         }
 960         delta_data = get_data_from_pack(delta_obj);
 961         base_data = get_base_data(base);
 962         result->obj = delta_obj;
 963         result->data = patch_delta(base_data, base->size,
 964                                    delta_data, delta_obj->size, &result->size);
 965         free(delta_data);
 966         if (!result->data)
 967                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 968         hash_object_file(result->data, result->size,
 969                          type_name(delta_obj->real_type), &delta_obj->idx.oid);
 970         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 971                     &delta_obj->idx.oid);
 972         counter_lock();
 973         nr_resolved_deltas++;
 974         counter_unlock();
 975 }
 976
 977 /*
 978  * Standard boolean compare-and-swap: atomically check whether "*type" is
 979  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 980  * and return false.
 981  */
 982 static int compare_and_swap_type(signed char *type,
 983                                  enum object_type want,
 984                                  enum object_type set)
 985 {
 986         enum object_type old;
 987
 988         type_cas_lock();
 989         old = *type;
 990         if (old == want)
 991                 *type = set;
 992         type_cas_unlock();
 993
 994         return old == want;
 995 }
 996
 997 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 998                                                   struct base_data *prev_base)
 999 {
1000         if (base->ref_last == -1 && base->ofs_last == -1) {
1001                 find_ref_delta_children(&base->obj->idx.oid,
1002                                         &base->ref_first, &base->ref_last,
1003                                         OBJ_REF_DELTA);
1004
1005                 find_ofs_delta_children(base->obj->idx.offset,
1006                                         &base->ofs_first, &base->ofs_last,
1007                                         OBJ_OFS_DELTA);
1008
1009                 if (base->ref_last == -1 && base->ofs_last == -1) {
1010                         free(base->data);
1011                         return NULL;
1012                 }
1013
1014                 link_base_data(prev_base, base);
1015         }
1016
1017         if (base->ref_first <= base->ref_last) {
1018                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1019                 struct base_data *result = alloc_base_data();
1020
1021                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1022                                            base->obj->real_type))
1023                         BUG("child->real_type != OBJ_REF_DELTA");
1024
1025                 resolve_delta(child, base, result);
1026                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1027                         free_base_data(base);
1028
1029                 base->ref_first++;
1030                 return result;
1031         }
1032
1033         if (base->ofs_first <= base->ofs_last) {
1034                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1035                 struct base_data *result = alloc_base_data();
1036
1037                 assert(child->real_type == OBJ_OFS_DELTA);
1038                 child->real_type = base->obj->real_type;
1039                 resolve_delta(child, base, result);
1040                 if (base->ofs_first == base->ofs_last)
1041                         free_base_data(base);
1042
1043                 base->ofs_first++;
1044                 return result;
1045         }
1046
1047         unlink_base_data(base);
1048         return NULL;
1049 }
1050
1051 static void find_unresolved_deltas(struct base_data *base)
1052 {
1053         struct base_data *new_base, *prev_base = NULL;
1054         for (;;) {
1055                 new_base = find_unresolved_deltas_1(base, prev_base);
1056
1057                 if (new_base) {
1058                         prev_base = base;
1059                         base = new_base;
1060                 } else {
1061                         free(base);
1062                         base = prev_base;
1063                         if (!base)
1064                                 return;
1065                         prev_base = base->base;
1066                 }
1067         }
1068 }
1069
1070 static int compare_ofs_delta_entry(const void *a, const void *b)
1071 {
1072         const struct ofs_delta_entry *delta_a = a;
1073         const struct ofs_delta_entry *delta_b = b;
1074
1075         return delta_a->offset < delta_b->offset ? -1 :
1076                delta_a->offset > delta_b->offset ?  1 :
1077                0;
1078 }
1079
1080 static int compare_ref_delta_entry(const void *a, const void *b)
1081 {
1082         const struct ref_delta_entry *delta_a = a;
1083         const struct ref_delta_entry *delta_b = b;
1084
1085         return oidcmp(&delta_a->oid, &delta_b->oid);
1086 }
1087
1088 static void resolve_base(struct object_entry *obj)
1089 {
1090         struct base_data *base_obj = alloc_base_data();
1091         base_obj->obj = obj;
1092         base_obj->data = NULL;
1093         find_unresolved_deltas(base_obj);
1094 }
1095
1096 #ifndef NO_PTHREADS
1097 static void *threaded_second_pass(void *data)
1098 {
1099         set_thread_data(data);
1100         for (;;) {
1101                 int i;
1102                 counter_lock();
1103                 display_progress(progress, nr_resolved_deltas);
1104                 counter_unlock();
1105                 work_lock();
1106                 while (nr_dispatched < nr_objects &&
1107                        is_delta_type(objects[nr_dispatched].type))
1108                         nr_dispatched++;
1109                 if (nr_dispatched >= nr_objects) {
1110                         work_unlock();
1111                         break;
1112                 }
1113                 i = nr_dispatched++;
1114                 work_unlock();
1115
1116                 resolve_base(&objects[i]);
1117         }
1118         return NULL;
1119 }
1120 #endif
1121
1122 /*
1123  * First pass:
1124  * - find locations of all objects;
1125  * - calculate SHA1 of all non-delta objects;
1126  * - remember base (SHA1 or offset) for all deltas.
1127  */
1128 static void parse_pack_objects(unsigned char *hash)
1129 {
1130         int i, nr_delays = 0;
1131         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1132         struct object_id ref_delta_oid;
1133         struct stat st;
1134
1135         if (verbose)
1136                 progress = start_progress(
1137                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1138                                 nr_objects);
1139         for (i = 0; i < nr_objects; i++) {
1140                 struct object_entry *obj = &objects[i];
1141                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1142                                               &ref_delta_oid,
1143                                               &obj->idx.oid);
1144                 obj->real_type = obj->type;
1145                 if (obj->type == OBJ_OFS_DELTA) {
1146                         nr_ofs_deltas++;
1147                         ofs_delta->obj_no = i;
1148                         ofs_delta++;
1149                 } else if (obj->type == OBJ_REF_DELTA) {
1150                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1151                         oidcpy(&ref_deltas[nr_ref_deltas].oid, &ref_delta_oid);
1152                         ref_deltas[nr_ref_deltas].obj_no = i;
1153                         nr_ref_deltas++;
1154                 } else if (!data) {
1155                         /* large blobs, check later */
1156                         obj->real_type = OBJ_BAD;
1157                         nr_delays++;
1158                 } else
1159                         sha1_object(data, NULL, obj->size, obj->type,
1160                                     &obj->idx.oid);
1161                 free(data);
1162                 display_progress(progress, i+1);
1163         }
1164         objects[i].idx.offset = consumed_bytes;
1165         stop_progress(&progress);
1166
1167         /* Check pack integrity */
1168         flush();
1169         the_hash_algo->final_fn(hash, &input_ctx);
1170         if (!hasheq(fill(the_hash_algo->rawsz), hash))
1171                 die(_("pack is corrupted (SHA1 mismatch)"));
1172         use(the_hash_algo->rawsz);
1173
1174         /* If input_fd is a file, we should have reached its end now. */
1175         if (fstat(input_fd, &st))
1176                 die_errno(_("cannot fstat packfile"));
1177         if (S_ISREG(st.st_mode) &&
1178                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1179                 die(_("pack has junk at the end"));
1180
1181         for (i = 0; i < nr_objects; i++) {
1182                 struct object_entry *obj = &objects[i];
1183                 if (obj->real_type != OBJ_BAD)
1184                         continue;
1185                 obj->real_type = obj->type;
1186                 sha1_object(NULL, obj, obj->size, obj->type,
1187                             &obj->idx.oid);
1188                 nr_delays--;
1189         }
1190         if (nr_delays)
1191                 die(_("confusion beyond insanity in parse_pack_objects()"));
1192 }
1193
1194 /*
1195  * Second pass:
1196  * - for all non-delta objects, look if it is used as a base for
1197  *   deltas;
1198  * - if used as a base, uncompress the object and apply all deltas,
1199  *   recursively checking if the resulting object is used as a base
1200  *   for some more deltas.
1201  */
1202 static void resolve_deltas(void)
1203 {
1204         int i;
1205
1206         if (!nr_ofs_deltas && !nr_ref_deltas)
1207                 return;
1208
1209         /* Sort deltas by base SHA1/offset for fast searching */
1210         QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1211         QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1212
1213         if (verbose || show_resolving_progress)
1214                 progress = start_progress(_("Resolving deltas"),
1215                                           nr_ref_deltas + nr_ofs_deltas);
1216
1217 #ifndef NO_PTHREADS
1218         nr_dispatched = 0;
1219         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1220                 init_thread();
1221                 for (i = 0; i < nr_threads; i++) {
1222                         int ret = pthread_create(&thread_data[i].thread, NULL,
1223                                                  threaded_second_pass, thread_data + i);
1224                         if (ret)
1225                                 die(_("unable to create thread: %s"),
1226                                     strerror(ret));
1227                 }
1228                 for (i = 0; i < nr_threads; i++)
1229                         pthread_join(thread_data[i].thread, NULL);
1230                 cleanup_thread();
1231                 return;
1232         }
1233 #endif
1234
1235         for (i = 0; i < nr_objects; i++) {
1236                 struct object_entry *obj = &objects[i];
1237
1238                 if (is_delta_type(obj->type))
1239                         continue;
1240                 resolve_base(obj);
1241                 display_progress(progress, nr_resolved_deltas);
1242         }
1243 }
1244
1245 /*
1246  * Third pass:
1247  * - append objects to convert thin pack to full pack if required
1248  * - write the final pack hash
1249  */
1250 static void fix_unresolved_deltas(struct hashfile *f);
1251 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_hash)
1252 {
1253         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1254                 stop_progress(&progress);
1255                 /* Flush remaining pack final hash. */
1256                 flush();
1257                 return;
1258         }
1259
1260         if (fix_thin_pack) {
1261                 struct hashfile *f;
1262                 unsigned char read_hash[GIT_MAX_RAWSZ], tail_hash[GIT_MAX_RAWSZ];
1263                 struct strbuf msg = STRBUF_INIT;
1264                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1265                 int nr_objects_initial = nr_objects;
1266                 if (nr_unresolved <= 0)
1267                         die(_("confusion beyond insanity"));
1268                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1269                 memset(objects + nr_objects + 1, 0,
1270                        nr_unresolved * sizeof(*objects));
1271                 f = hashfd(output_fd, curr_pack);
1272                 fix_unresolved_deltas(f);
1273                 strbuf_addf(&msg, Q_("completed with %d local object",
1274                                      "completed with %d local objects",
1275                                      nr_objects - nr_objects_initial),
1276                             nr_objects - nr_objects_initial);
1277                 stop_progress_msg(&progress, msg.buf);
1278                 strbuf_release(&msg);
1279                 finalize_hashfile(f, tail_hash, 0);
1280                 hashcpy(read_hash, pack_hash);
1281                 fixup_pack_header_footer(output_fd, pack_hash,
1282                                          curr_pack, nr_objects,
1283                                          read_hash, consumed_bytes-the_hash_algo->rawsz);
1284                 if (!hasheq(read_hash, tail_hash))
1285                         die(_("Unexpected tail checksum for %s "
1286                               "(disk corruption?)"), curr_pack);
1287         }
1288         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1289                 die(Q_("pack has %d unresolved delta",
1290                        "pack has %d unresolved deltas",
1291                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1292                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1293 }
1294
1295 static int write_compressed(struct hashfile *f, void *in, unsigned int size)
1296 {
1297         git_zstream stream;
1298         int status;
1299         unsigned char outbuf[4096];
1300
1301         git_deflate_init(&stream, zlib_compression_level);
1302         stream.next_in = in;
1303         stream.avail_in = size;
1304
1305         do {
1306                 stream.next_out = outbuf;
1307                 stream.avail_out = sizeof(outbuf);
1308                 status = git_deflate(&stream, Z_FINISH);
1309                 hashwrite(f, outbuf, sizeof(outbuf) - stream.avail_out);
1310         } while (status == Z_OK);
1311
1312         if (status != Z_STREAM_END)
1313                 die(_("unable to deflate appended object (%d)"), status);
1314         size = stream.total_out;
1315         git_deflate_end(&stream);
1316         return size;
1317 }
1318
1319 static struct object_entry *append_obj_to_pack(struct hashfile *f,
1320                                const unsigned char *sha1, void *buf,
1321                                unsigned long size, enum object_type type)
1322 {
1323         struct object_entry *obj = &objects[nr_objects++];
1324         unsigned char header[10];
1325         unsigned long s = size;
1326         int n = 0;
1327         unsigned char c = (type << 4) | (s & 15);
1328         s >>= 4;
1329         while (s) {
1330                 header[n++] = c | 0x80;
1331                 c = s & 0x7f;
1332                 s >>= 7;
1333         }
1334         header[n++] = c;
1335         crc32_begin(f);
1336         hashwrite(f, header, n);
1337         obj[0].size = size;
1338         obj[0].hdr_size = n;
1339         obj[0].type = type;
1340         obj[0].real_type = type;
1341         obj[1].idx.offset = obj[0].idx.offset + n;
1342         obj[1].idx.offset += write_compressed(f, buf, size);
1343         obj[0].idx.crc32 = crc32_end(f);
1344         hashflush(f);
1345         hashcpy(obj->idx.oid.hash, sha1);
1346         return obj;
1347 }
1348
1349 static int delta_pos_compare(const void *_a, const void *_b)
1350 {
1351         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1352         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1353         return a->obj_no - b->obj_no;
1354 }
1355
1356 static void fix_unresolved_deltas(struct hashfile *f)
1357 {
1358         struct ref_delta_entry **sorted_by_pos;
1359         int i;
1360
1361         /*
1362          * Since many unresolved deltas may well be themselves base objects
1363          * for more unresolved deltas, we really want to include the
1364          * smallest number of base objects that would cover as much delta
1365          * as possible by picking the
1366          * trunc deltas first, allowing for other deltas to resolve without
1367          * additional base objects.  Since most base objects are to be found
1368          * before deltas depending on them, a good heuristic is to start
1369          * resolving deltas in the same order as their position in the pack.
1370          */
1371         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1372         for (i = 0; i < nr_ref_deltas; i++)
1373                 sorted_by_pos[i] = &ref_deltas[i];
1374         QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
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(&d->oid, base_obj->data,
1389                                 base_obj->size, type_name(type)))
1390                         die(_("local object %s is corrupt"), oid_to_hex(&d->oid));
1391                 base_obj->obj = append_obj_to_pack(f, d->oid.hash,
1392                                         base_obj->data, base_obj->size, type);
1393                 find_unresolved_deltas(base_obj);
1394                 display_progress(progress, nr_resolved_deltas);
1395         }
1396         free(sorted_by_pos);
1397 }
1398
1399 static const char *derive_filename(const char *pack_name, const char *suffix,
1400                                    struct strbuf *buf)
1401 {
1402         size_t len;
1403         if (!strip_suffix(pack_name, ".pack", &len))
1404                 die(_("packfile name '%s' does not end with '.pack'"),
1405                     pack_name);
1406         strbuf_add(buf, pack_name, len);
1407         strbuf_addch(buf, '.');
1408         strbuf_addstr(buf, suffix);
1409         return buf->buf;
1410 }
1411
1412 static void write_special_file(const char *suffix, const char *msg,
1413                                const char *pack_name, const unsigned char *hash,
1414                                const char **report)
1415 {
1416         struct strbuf name_buf = STRBUF_INIT;
1417         const char *filename;
1418         int fd;
1419         int msg_len = strlen(msg);
1420
1421         if (pack_name)
1422                 filename = derive_filename(pack_name, suffix, &name_buf);
1423         else
1424                 filename = odb_pack_name(&name_buf, hash, suffix);
1425
1426         fd = odb_pack_keep(filename);
1427         if (fd < 0) {
1428                 if (errno != EEXIST)
1429                         die_errno(_("cannot write %s file '%s'"),
1430                                   suffix, filename);
1431         } else {
1432                 if (msg_len > 0) {
1433                         write_or_die(fd, msg, msg_len);
1434                         write_or_die(fd, "\n", 1);
1435                 }
1436                 if (close(fd) != 0)
1437                         die_errno(_("cannot close written %s file '%s'"),
1438                                   suffix, filename);
1439                 if (report)
1440                         *report = suffix;
1441         }
1442         strbuf_release(&name_buf);
1443 }
1444
1445 static void final(const char *final_pack_name, const char *curr_pack_name,
1446                   const char *final_index_name, const char *curr_index_name,
1447                   const char *keep_msg, const char *promisor_msg,
1448                   unsigned char *hash)
1449 {
1450         const char *report = "pack";
1451         struct strbuf pack_name = STRBUF_INIT;
1452         struct strbuf index_name = STRBUF_INIT;
1453         int err;
1454
1455         if (!from_stdin) {
1456                 close(input_fd);
1457         } else {
1458                 fsync_or_die(output_fd, curr_pack_name);
1459                 err = close(output_fd);
1460                 if (err)
1461                         die_errno(_("error while closing pack file"));
1462         }
1463
1464         if (keep_msg)
1465                 write_special_file("keep", keep_msg, final_pack_name, hash,
1466                                    &report);
1467         if (promisor_msg)
1468                 write_special_file("promisor", promisor_msg, final_pack_name,
1469                                    hash, NULL);
1470
1471         if (final_pack_name != curr_pack_name) {
1472                 if (!final_pack_name)
1473                         final_pack_name = odb_pack_name(&pack_name, hash, "pack");
1474                 if (finalize_object_file(curr_pack_name, final_pack_name))
1475                         die(_("cannot store pack file"));
1476         } else if (from_stdin)
1477                 chmod(final_pack_name, 0444);
1478
1479         if (final_index_name != curr_index_name) {
1480                 if (!final_index_name)
1481                         final_index_name = odb_pack_name(&index_name, hash, "idx");
1482                 if (finalize_object_file(curr_index_name, final_index_name))
1483                         die(_("cannot store index file"));
1484         } else
1485                 chmod(final_index_name, 0444);
1486
1487         if (do_fsck_object) {
1488                 struct packed_git *p;
1489                 p = add_packed_git(final_index_name, strlen(final_index_name), 0);
1490                 if (p)
1491                         install_packed_git(the_repository, p);
1492         }
1493
1494         if (!from_stdin) {
1495                 printf("%s\n", sha1_to_hex(hash));
1496         } else {
1497                 struct strbuf buf = STRBUF_INIT;
1498
1499                 strbuf_addf(&buf, "%s\t%s\n", report, sha1_to_hex(hash));
1500                 write_or_die(1, buf.buf, buf.len);
1501                 strbuf_release(&buf);
1502
1503                 /*
1504                  * Let's just mimic git-unpack-objects here and write
1505                  * the last part of the input buffer to stdout.
1506                  */
1507                 while (input_len) {
1508                         err = xwrite(1, input_buffer + input_offset, input_len);
1509                         if (err <= 0)
1510                                 break;
1511                         input_len -= err;
1512                         input_offset += err;
1513                 }
1514         }
1515
1516         strbuf_release(&index_name);
1517         strbuf_release(&pack_name);
1518 }
1519
1520 static int git_index_pack_config(const char *k, const char *v, void *cb)
1521 {
1522         struct pack_idx_option *opts = cb;
1523
1524         if (!strcmp(k, "pack.indexversion")) {
1525                 opts->version = git_config_int(k, v);
1526                 if (opts->version > 2)
1527                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1528                 return 0;
1529         }
1530         if (!strcmp(k, "pack.threads")) {
1531                 nr_threads = git_config_int(k, v);
1532                 if (nr_threads < 0)
1533                         die(_("invalid number of threads specified (%d)"),
1534                             nr_threads);
1535 #ifdef NO_PTHREADS
1536                 if (nr_threads != 1)
1537                         warning(_("no threads support, ignoring %s"), k);
1538                 nr_threads = 1;
1539 #endif
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, since it only
1677          * accesses the repo to do hash collision checks
1678          */
1679         fetch_if_missing = 0;
1680
1681         if (argc == 2 && !strcmp(argv[1], "-h"))
1682                 usage(index_pack_usage);
1683
1684         read_replace_refs = 0;
1685         fsck_options.walk = mark_link;
1686
1687         reset_pack_idx_option(&opts);
1688         git_config(git_index_pack_config, &opts);
1689         if (prefix && chdir(prefix))
1690                 die(_("Cannot come back to cwd"));
1691
1692         for (i = 1; i < argc; i++) {
1693                 const char *arg = argv[i];
1694
1695                 if (*arg == '-') {
1696                         if (!strcmp(arg, "--stdin")) {
1697                                 from_stdin = 1;
1698                         } else if (!strcmp(arg, "--fix-thin")) {
1699                                 fix_thin_pack = 1;
1700                         } else if (skip_to_optional_arg(arg, "--strict", &arg)) {
1701                                 strict = 1;
1702                                 do_fsck_object = 1;
1703                                 fsck_set_msg_types(&fsck_options, arg);
1704                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1705                                 strict = 1;
1706                                 check_self_contained_and_connected = 1;
1707                         } else if (!strcmp(arg, "--fsck-objects")) {
1708                                 do_fsck_object = 1;
1709                         } else if (!strcmp(arg, "--verify")) {
1710                                 verify = 1;
1711                         } else if (!strcmp(arg, "--verify-stat")) {
1712                                 verify = 1;
1713                                 show_stat = 1;
1714                         } else if (!strcmp(arg, "--verify-stat-only")) {
1715                                 verify = 1;
1716                                 show_stat = 1;
1717                                 stat_only = 1;
1718                         } else if (skip_to_optional_arg(arg, "--keep", &keep_msg)) {
1719                                 ; /* nothing to do */
1720                         } else if (skip_to_optional_arg(arg, "--promisor", &promisor_msg)) {
1721                                 ; /* already parsed */
1722                         } else if (starts_with(arg, "--threads=")) {
1723                                 char *end;
1724                                 nr_threads = strtoul(arg+10, &end, 0);
1725                                 if (!arg[10] || *end || nr_threads < 0)
1726                                         usage(index_pack_usage);
1727 #ifdef NO_PTHREADS
1728                                 if (nr_threads != 1)
1729                                         warning(_("no threads support, "
1730                                                   "ignoring %s"), arg);
1731                                 nr_threads = 1;
1732 #endif
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 #ifndef NO_PTHREADS
1796         if (!nr_threads) {
1797                 nr_threads = online_cpus();
1798                 /* An experiment showed that more threads does not mean faster */
1799                 if (nr_threads > 3)
1800                         nr_threads = 3;
1801         }
1802 #endif
1803
1804         curr_pack = open_pack_file(pack_name);
1805         parse_pack_header();
1806         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1807         if (show_stat)
1808                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1809         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1810         parse_pack_objects(pack_hash);
1811         if (report_end_of_input)
1812                 write_in_full(2, "\0", 1);
1813         resolve_deltas();
1814         conclude_pack(fix_thin_pack, curr_pack, pack_hash);
1815         free(ofs_deltas);
1816         free(ref_deltas);
1817         if (strict)
1818                 foreign_nr = check_objects();
1819
1820         if (show_stat)
1821                 show_pack_info(stat_only);
1822
1823         ALLOC_ARRAY(idx_objects, nr_objects);
1824         for (i = 0; i < nr_objects; i++)
1825                 idx_objects[i] = &objects[i].idx;
1826         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_hash);
1827         free(idx_objects);
1828
1829         if (!verify)
1830                 final(pack_name, curr_pack,
1831                       index_name, curr_index,
1832                       keep_msg, promisor_msg,
1833                       pack_hash);
1834         else
1835                 close(input_fd);
1836
1837         if (do_fsck_object && fsck_finish(&fsck_options))
1838                 die(_("fsck error in pack objects"));
1839
1840         free(objects);
1841         strbuf_release(&index_name_buf);
1842         if (pack_name == NULL)
1843                 free((void *) curr_pack);
1844         if (index_name == NULL)
1845                 free((void *) curr_index);
1846
1847         /*
1848          * Let the caller know this pack is not self contained
1849          */
1850         if (check_self_contained_and_connected && foreign_nr)
1851                 return 1;
1852
1853         return 0;
1854 }