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         pthread_t thread;
  46         struct base_data *base_cache;
  47         size_t base_cache_used;
  48         int pack_fd;
  49 };
  50
  51 /* Remember to update object flag allocation in object.h */
  52 #define FLAG_LINK (1u<<20)
  53 #define FLAG_CHECKED (1u<<21)
  54
  55 struct ofs_delta_entry {
  56         off_t offset;
  57         int obj_no;
  58 };
  59
  60 struct ref_delta_entry {
  61         struct object_id oid;
  62         int obj_no;
  63 };
  64
  65 static struct object_entry *objects;
  66 static struct object_stat *obj_stat;
  67 static struct ofs_delta_entry *ofs_deltas;
  68 static struct ref_delta_entry *ref_deltas;
  69 static struct thread_local nothread_data;
  70 static int nr_objects;
  71 static int nr_ofs_deltas;
  72 static int nr_ref_deltas;
  73 static int ref_deltas_alloc;
  74 static int nr_resolved_deltas;
  75 static int nr_threads;
  76
  77 static int from_stdin;
  78 static int strict;
  79 static int do_fsck_object;
  80 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  81 static int verbose;
  82 static int show_resolving_progress;
  83 static int show_stat;
  84 static int check_self_contained_and_connected;
  85
  86 static struct progress *progress;
  87
  88 /* We always read in 4kB chunks. */
  89 static unsigned char input_buffer[4096];
  90 static unsigned int input_offset, input_len;
  91 static off_t consumed_bytes;
  92 static off_t max_input_size;
  93 static unsigned deepest_delta;
  94 static git_hash_ctx input_ctx;
  95 static uint32_t input_crc32;
  96 static int input_fd, output_fd;
  97 static const char *curr_pack;
  98
  99 static struct thread_local *thread_data;
 100 static int nr_dispatched;
 101 static int threads_active;
 102
 103 static pthread_mutex_t read_mutex;
 104 #define read_lock()             lock_mutex(&read_mutex)
 105 #define read_unlock()           unlock_mutex(&read_mutex)
 106
 107 static pthread_mutex_t counter_mutex;
 108 #define counter_lock()          lock_mutex(&counter_mutex)
 109 #define counter_unlock()        unlock_mutex(&counter_mutex)
 110
 111 static pthread_mutex_t work_mutex;
 112 #define work_lock()             lock_mutex(&work_mutex)
 113 #define work_unlock()           unlock_mutex(&work_mutex)
 114
 115 static pthread_mutex_t deepest_delta_mutex;
 116 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 117 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 118
 119 static pthread_mutex_t type_cas_mutex;
 120 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 121 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 122
 123 static pthread_key_t key;
 124
 125 static inline void lock_mutex(pthread_mutex_t *mutex)
 126 {
 127         if (threads_active)
 128                 pthread_mutex_lock(mutex);
 129 }
 130
 131 static inline void unlock_mutex(pthread_mutex_t *mutex)
 132 {
 133         if (threads_active)
 134                 pthread_mutex_unlock(mutex);
 135 }
 136
 137 /*
 138  * Mutex and conditional variable can't be statically-initialized on Windows.
 139  */
 140 static void init_thread(void)
 141 {
 142         int i;
 143         init_recursive_mutex(&read_mutex);
 144         pthread_mutex_init(&counter_mutex, NULL);
 145         pthread_mutex_init(&work_mutex, NULL);
 146         pthread_mutex_init(&type_cas_mutex, NULL);
 147         if (show_stat)
 148                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 149         pthread_key_create(&key, NULL);
 150         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 151         for (i = 0; i < nr_threads; i++) {
 152                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 153                 if (thread_data[i].pack_fd == -1)
 154                         die_errno(_("unable to open %s"), curr_pack);
 155         }
 156
 157         threads_active = 1;
 158 }
 159
 160 static void cleanup_thread(void)
 161 {
 162         int i;
 163         if (!threads_active)
 164                 return;
 165         threads_active = 0;
 166         pthread_mutex_destroy(&read_mutex);
 167         pthread_mutex_destroy(&counter_mutex);
 168         pthread_mutex_destroy(&work_mutex);
 169         pthread_mutex_destroy(&type_cas_mutex);
 170         if (show_stat)
 171                 pthread_mutex_destroy(&deepest_delta_mutex);
 172         for (i = 0; i < nr_threads; i++)
 173                 close(thread_data[i].pack_fd);
 174         pthread_key_delete(key);
 175         free(thread_data);
 176 }
 177
 178 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 179 {
 180         if (!obj)
 181                 return -1;
 182
 183         if (type != OBJ_ANY && obj->type != type)
 184                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 185
 186         obj->flags |= FLAG_LINK;
 187         return 0;
 188 }
 189
 190 /* The content of each linked object must have been checked
 191    or it must be already present in the object database */
 192 static unsigned check_object(struct object *obj)
 193 {
 194         if (!obj)
 195                 return 0;
 196
 197         if (!(obj->flags & FLAG_LINK))
 198                 return 0;
 199
 200         if (!(obj->flags & FLAG_CHECKED)) {
 201                 unsigned long size;
 202                 int type = oid_object_info(the_repository, &obj->oid, &size);
 203                 if (type <= 0)
 204                         die(_("did not receive expected object %s"),
 205                               oid_to_hex(&obj->oid));
 206                 if (type != obj->type)
 207                         die(_("object %s: expected type %s, found %s"),
 208                             oid_to_hex(&obj->oid),
 209                             type_name(obj->type), type_name(type));
 210                 obj->flags |= FLAG_CHECKED;
 211                 return 1;
 212         }
 213
 214         return 0;
 215 }
 216
 217 static unsigned check_objects(void)
 218 {
 219         unsigned i, max, foreign_nr = 0;
 220
 221         max = get_max_object_index();
 222         for (i = 0; i < max; i++)
 223                 foreign_nr += check_object(get_indexed_object(i));
 224         return foreign_nr;
 225 }
 226
 227
 228 /* Discard current buffer used content. */
 229 static void flush(void)
 230 {
 231         if (input_offset) {
 232                 if (output_fd >= 0)
 233                         write_or_die(output_fd, input_buffer, input_offset);
 234                 the_hash_algo->update_fn(&input_ctx, input_buffer, input_offset);
 235                 memmove(input_buffer, input_buffer + input_offset, input_len);
 236                 input_offset = 0;
 237         }
 238 }
 239
 240 /*
 241  * Make sure at least "min" bytes are available in the buffer, and
 242  * return the pointer to the buffer.
 243  */
 244 static void *fill(int min)
 245 {
 246         if (min <= input_len)
 247                 return input_buffer + input_offset;
 248         if (min > sizeof(input_buffer))
 249                 die(Q_("cannot fill %d byte",
 250                        "cannot fill %d bytes",
 251                        min),
 252                     min);
 253         flush();
 254         do {
 255                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 256                                 sizeof(input_buffer) - input_len);
 257                 if (ret <= 0) {
 258                         if (!ret)
 259                                 die(_("early EOF"));
 260                         die_errno(_("read error on input"));
 261                 }
 262                 input_len += ret;
 263                 if (from_stdin)
 264                         display_throughput(progress, consumed_bytes + input_len);
 265         } while (input_len < min);
 266         return input_buffer;
 267 }
 268
 269 static void use(int bytes)
 270 {
 271         if (bytes > input_len)
 272                 die(_("used more bytes than were available"));
 273         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 274         input_len -= bytes;
 275         input_offset += bytes;
 276
 277         /* make sure off_t is sufficiently large not to wrap */
 278         if (signed_add_overflows(consumed_bytes, bytes))
 279                 die(_("pack too large for current definition of off_t"));
 280         consumed_bytes += bytes;
 281         if (max_input_size && consumed_bytes > max_input_size)
 282                 die(_("pack exceeds maximum allowed size"));
 283 }
 284
 285 static const char *open_pack_file(const char *pack_name)
 286 {
 287         if (from_stdin) {
 288                 input_fd = 0;
 289                 if (!pack_name) {
 290                         struct strbuf tmp_file = STRBUF_INIT;
 291                         output_fd = odb_mkstemp(&tmp_file,
 292                                                 "pack/tmp_pack_XXXXXX");
 293                         pack_name = strbuf_detach(&tmp_file, NULL);
 294                 } else {
 295                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 296                         if (output_fd < 0)
 297                                 die_errno(_("unable to create '%s'"), pack_name);
 298                 }
 299                 nothread_data.pack_fd = output_fd;
 300         } else {
 301                 input_fd = open(pack_name, O_RDONLY);
 302                 if (input_fd < 0)
 303                         die_errno(_("cannot open packfile '%s'"), pack_name);
 304                 output_fd = -1;
 305                 nothread_data.pack_fd = input_fd;
 306         }
 307         the_hash_algo->init_fn(&input_ctx);
 308         return pack_name;
 309 }
 310
 311 static void parse_pack_header(void)
 312 {
 313         struct pack_header *hdr = fill(sizeof(struct pack_header));
 314
 315         /* Header consistency check */
 316         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 317                 die(_("pack signature mismatch"));
 318         if (!pack_version_ok(hdr->hdr_version))
 319                 die(_("pack version %"PRIu32" unsupported"),
 320                         ntohl(hdr->hdr_version));
 321
 322         nr_objects = ntohl(hdr->hdr_entries);
 323         use(sizeof(struct pack_header));
 324 }
 325
 326 static NORETURN void bad_object(off_t offset, const char *format,
 327                        ...) __attribute__((format (printf, 2, 3)));
 328
 329 static NORETURN void bad_object(off_t offset, const char *format, ...)
 330 {
 331         va_list params;
 332         char buf[1024];
 333
 334         va_start(params, format);
 335         vsnprintf(buf, sizeof(buf), format, params);
 336         va_end(params);
 337         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 338             (uintmax_t)offset, buf);
 339 }
 340
 341 static inline struct thread_local *get_thread_data(void)
 342 {
 343         if (HAVE_THREADS) {
 344                 if (threads_active)
 345                         return pthread_getspecific(key);
 346                 assert(!threads_active &&
 347                        "This should only be reached when all threads are gone");
 348         }
 349         return &nothread_data;
 350 }
 351
 352 static void set_thread_data(struct thread_local *data)
 353 {
 354         if (threads_active)
 355                 pthread_setspecific(key, data);
 356 }
 357
 358 static struct base_data *alloc_base_data(void)
 359 {
 360         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 361         base->ref_last = -1;
 362         base->ofs_last = -1;
 363         return base;
 364 }
 365
 366 static void free_base_data(struct base_data *c)
 367 {
 368         if (c->data) {
 369                 FREE_AND_NULL(c->data);
 370                 get_thread_data()->base_cache_used -= c->size;
 371         }
 372 }
 373
 374 static void prune_base_data(struct base_data *retain)
 375 {
 376         struct base_data *b;
 377         struct thread_local *data = get_thread_data();
 378         for (b = data->base_cache;
 379              data->base_cache_used > delta_base_cache_limit && b;
 380              b = b->child) {
 381                 if (b->data && b != retain)
 382                         free_base_data(b);
 383         }
 384 }
 385
 386 static void link_base_data(struct base_data *base, struct base_data *c)
 387 {
 388         if (base)
 389                 base->child = c;
 390         else
 391                 get_thread_data()->base_cache = c;
 392
 393         c->base = base;
 394         c->child = NULL;
 395         if (c->data)
 396                 get_thread_data()->base_cache_used += c->size;
 397         prune_base_data(c);
 398 }
 399
 400 static void unlink_base_data(struct base_data *c)
 401 {
 402         struct base_data *base = c->base;
 403         if (base)
 404                 base->child = NULL;
 405         else
 406                 get_thread_data()->base_cache = NULL;
 407         free_base_data(c);
 408 }
 409
 410 static int is_delta_type(enum object_type type)
 411 {
 412         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 413 }
 414
 415 static void *unpack_entry_data(off_t offset, unsigned long size,
 416                                enum object_type type, struct object_id *oid)
 417 {
 418         static char fixed_buf[8192];
 419         int status;
 420         git_zstream stream;
 421         void *buf;
 422         git_hash_ctx c;
 423         char hdr[32];
 424         int hdrlen;
 425
 426         if (!is_delta_type(type)) {
 427                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %"PRIuMAX,
 428                                    type_name(type),(uintmax_t)size) + 1;
 429                 the_hash_algo->init_fn(&c);
 430                 the_hash_algo->update_fn(&c, hdr, hdrlen);
 431         } else
 432                 oid = NULL;
 433         if (type == OBJ_BLOB && size > big_file_threshold)
 434                 buf = fixed_buf;
 435         else
 436                 buf = xmallocz(size);
 437
 438         memset(&stream, 0, sizeof(stream));
 439         git_inflate_init(&stream);
 440         stream.next_out = buf;
 441         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 442
 443         do {
 444                 unsigned char *last_out = stream.next_out;
 445                 stream.next_in = fill(1);
 446                 stream.avail_in = input_len;
 447                 status = git_inflate(&stream, 0);
 448                 use(input_len - stream.avail_in);
 449                 if (oid)
 450                         the_hash_algo->update_fn(&c, last_out, stream.next_out - last_out);
 451                 if (buf == fixed_buf) {
 452                         stream.next_out = buf;
 453                         stream.avail_out = sizeof(fixed_buf);
 454                 }
 455         } while (status == Z_OK);
 456         if (stream.total_out != size || status != Z_STREAM_END)
 457                 bad_object(offset, _("inflate returned %d"), status);
 458         git_inflate_end(&stream);
 459         if (oid)
 460                 the_hash_algo->final_fn(oid->hash, &c);
 461         return buf == fixed_buf ? NULL : buf;
 462 }
 463
 464 static void *unpack_raw_entry(struct object_entry *obj,
 465                               off_t *ofs_offset,
 466                               struct object_id *ref_oid,
 467                               struct object_id *oid)
 468 {
 469         unsigned char *p;
 470         unsigned long size, c;
 471         off_t base_offset;
 472         unsigned shift;
 473         void *data;
 474
 475         obj->idx.offset = consumed_bytes;
 476         input_crc32 = crc32(0, NULL, 0);
 477
 478         p = fill(1);
 479         c = *p;
 480         use(1);
 481         obj->type = (c >> 4) & 7;
 482         size = (c & 15);
 483         shift = 4;
 484         while (c & 0x80) {
 485                 p = fill(1);
 486                 c = *p;
 487                 use(1);
 488                 size += (c & 0x7f) << shift;
 489                 shift += 7;
 490         }
 491         obj->size = size;
 492
 493         switch (obj->type) {
 494         case OBJ_REF_DELTA:
 495                 hashcpy(ref_oid->hash, fill(the_hash_algo->rawsz));
 496                 use(the_hash_algo->rawsz);
 497                 break;
 498         case OBJ_OFS_DELTA:
 499                 p = fill(1);
 500                 c = *p;
 501                 use(1);
 502                 base_offset = c & 127;
 503                 while (c & 128) {
 504                         base_offset += 1;
 505                         if (!base_offset || MSB(base_offset, 7))
 506                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 507                         p = fill(1);
 508                         c = *p;
 509                         use(1);
 510                         base_offset = (base_offset << 7) + (c & 127);
 511                 }
 512                 *ofs_offset = obj->idx.offset - base_offset;
 513                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 514                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 515                 break;
 516         case OBJ_COMMIT:
 517         case OBJ_TREE:
 518         case OBJ_BLOB:
 519         case OBJ_TAG:
 520                 break;
 521         default:
 522                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 523         }
 524         obj->hdr_size = consumed_bytes - obj->idx.offset;
 525
 526         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, oid);
 527         obj->idx.crc32 = input_crc32;
 528         return data;
 529 }
 530
 531 static void *unpack_data(struct object_entry *obj,
 532                          int (*consume)(const unsigned char *, unsigned long, void *),
 533                          void *cb_data)
 534 {
 535         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 536         off_t len = obj[1].idx.offset - from;
 537         unsigned char *data, *inbuf;
 538         git_zstream stream;
 539         int status;
 540
 541         data = xmallocz(consume ? 64*1024 : obj->size);
 542         inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
 543
 544         memset(&stream, 0, sizeof(stream));
 545         git_inflate_init(&stream);
 546         stream.next_out = data;
 547         stream.avail_out = consume ? 64*1024 : obj->size;
 548
 549         do {
 550                 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
 551                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 552                 if (n < 0)
 553                         die_errno(_("cannot pread pack file"));
 554                 if (!n)
 555                         die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
 556                                "premature end of pack file, %"PRIuMAX" bytes missing",
 557                                (unsigned int)len),
 558                             (uintmax_t)len);
 559                 from += n;
 560                 len -= n;
 561                 stream.next_in = inbuf;
 562                 stream.avail_in = n;
 563                 if (!consume)
 564                         status = git_inflate(&stream, 0);
 565                 else {
 566                         do {
 567                                 status = git_inflate(&stream, 0);
 568                                 if (consume(data, stream.next_out - data, cb_data)) {
 569                                         free(inbuf);
 570                                         free(data);
 571                                         return NULL;
 572                                 }
 573                                 stream.next_out = data;
 574                                 stream.avail_out = 64*1024;
 575                         } while (status == Z_OK && stream.avail_in);
 576                 }
 577         } while (len && status == Z_OK && !stream.avail_in);
 578
 579         /* This has been inflated OK when first encountered, so... */
 580         if (status != Z_STREAM_END || stream.total_out != obj->size)
 581                 die(_("serious inflate inconsistency"));
 582
 583         git_inflate_end(&stream);
 584         free(inbuf);
 585         if (consume) {
 586                 FREE_AND_NULL(data);
 587         }
 588         return data;
 589 }
 590
 591 static void *get_data_from_pack(struct object_entry *obj)
 592 {
 593         return unpack_data(obj, NULL, NULL);
 594 }
 595
 596 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 597                                    enum object_type type1,
 598                                    enum object_type type2)
 599 {
 600         int cmp = type1 - type2;
 601         if (cmp)
 602                 return cmp;
 603         return offset1 < offset2 ? -1 :
 604                offset1 > offset2 ?  1 :
 605                0;
 606 }
 607
 608 static int find_ofs_delta(const off_t offset, enum object_type type)
 609 {
 610         int first = 0, last = nr_ofs_deltas;
 611
 612         while (first < last) {
 613                 int next = first + (last - first) / 2;
 614                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 615                 int cmp;
 616
 617                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 618                                               type, objects[delta->obj_no].type);
 619                 if (!cmp)
 620                         return next;
 621                 if (cmp < 0) {
 622                         last = next;
 623                         continue;
 624                 }
 625                 first = next+1;
 626         }
 627         return -first-1;
 628 }
 629
 630 static void find_ofs_delta_children(off_t offset,
 631                                     int *first_index, int *last_index,
 632                                     enum object_type type)
 633 {
 634         int first = find_ofs_delta(offset, type);
 635         int last = first;
 636         int end = nr_ofs_deltas - 1;
 637
 638         if (first < 0) {
 639                 *first_index = 0;
 640                 *last_index = -1;
 641                 return;
 642         }
 643         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 644                 --first;
 645         while (last < end && ofs_deltas[last + 1].offset == offset)
 646                 ++last;
 647         *first_index = first;
 648         *last_index = last;
 649 }
 650
 651 static int compare_ref_delta_bases(const struct object_id *oid1,
 652                                    const struct object_id *oid2,
 653                                    enum object_type type1,
 654                                    enum object_type type2)
 655 {
 656         int cmp = type1 - type2;
 657         if (cmp)
 658                 return cmp;
 659         return oidcmp(oid1, oid2);
 660 }
 661
 662 static int find_ref_delta(const struct object_id *oid, enum object_type type)
 663 {
 664         int first = 0, last = nr_ref_deltas;
 665
 666         while (first < last) {
 667                 int next = first + (last - first) / 2;
 668                 struct ref_delta_entry *delta = &ref_deltas[next];
 669                 int cmp;
 670
 671                 cmp = compare_ref_delta_bases(oid, &delta->oid,
 672                                               type, objects[delta->obj_no].type);
 673                 if (!cmp)
 674                         return next;
 675                 if (cmp < 0) {
 676                         last = next;
 677                         continue;
 678                 }
 679                 first = next+1;
 680         }
 681         return -first-1;
 682 }
 683
 684 static void find_ref_delta_children(const struct object_id *oid,
 685                                     int *first_index, int *last_index,
 686                                     enum object_type type)
 687 {
 688         int first = find_ref_delta(oid, type);
 689         int last = first;
 690         int end = nr_ref_deltas - 1;
 691
 692         if (first < 0) {
 693                 *first_index = 0;
 694                 *last_index = -1;
 695                 return;
 696         }
 697         while (first > 0 && oideq(&ref_deltas[first - 1].oid, oid))
 698                 --first;
 699         while (last < end && oideq(&ref_deltas[last + 1].oid, oid))
 700                 ++last;
 701         *first_index = first;
 702         *last_index = last;
 703 }
 704
 705 struct compare_data {
 706         struct object_entry *entry;
 707         struct git_istream *st;
 708         unsigned char *buf;
 709         unsigned long buf_size;
 710 };
 711
 712 static int compare_objects(const unsigned char *buf, unsigned long size,
 713                            void *cb_data)
 714 {
 715         struct compare_data *data = cb_data;
 716
 717         if (data->buf_size < size) {
 718                 free(data->buf);
 719                 data->buf = xmalloc(size);
 720                 data->buf_size = size;
 721         }
 722
 723         while (size) {
 724                 ssize_t len = read_istream(data->st, data->buf, size);
 725                 if (len == 0)
 726                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 727                             oid_to_hex(&data->entry->idx.oid));
 728                 if (len < 0)
 729                         die(_("unable to read %s"),
 730                             oid_to_hex(&data->entry->idx.oid));
 731                 if (memcmp(buf, data->buf, len))
 732                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 733                             oid_to_hex(&data->entry->idx.oid));
 734                 size -= len;
 735                 buf += len;
 736         }
 737         return 0;
 738 }
 739
 740 static int check_collison(struct object_entry *entry)
 741 {
 742         struct compare_data data;
 743         enum object_type type;
 744         unsigned long size;
 745
 746         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 747                 return -1;
 748
 749         memset(&data, 0, sizeof(data));
 750         data.entry = entry;
 751         data.st = open_istream(&entry->idx.oid, &type, &size, NULL);
 752         if (!data.st)
 753                 return -1;
 754         if (size != entry->size || type != entry->type)
 755                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 756                     oid_to_hex(&entry->idx.oid));
 757         unpack_data(entry, compare_objects, &data);
 758         close_istream(data.st);
 759         free(data.buf);
 760         return 0;
 761 }
 762
 763 static void sha1_object(const void *data, struct object_entry *obj_entry,
 764                         unsigned long size, enum object_type type,
 765                         const struct object_id *oid)
 766 {
 767         void *new_data = NULL;
 768         int collision_test_needed = 0;
 769
 770         assert(data || obj_entry);
 771
 772         if (startup_info->have_repository) {
 773                 read_lock();
 774                 collision_test_needed =
 775                         has_object_file_with_flags(oid, OBJECT_INFO_QUICK);
 776                 read_unlock();
 777         }
 778
 779         if (collision_test_needed && !data) {
 780                 read_lock();
 781                 if (!check_collison(obj_entry))
 782                         collision_test_needed = 0;
 783                 read_unlock();
 784         }
 785         if (collision_test_needed) {
 786                 void *has_data;
 787                 enum object_type has_type;
 788                 unsigned long has_size;
 789                 read_lock();
 790                 has_type = oid_object_info(the_repository, oid, &has_size);
 791                 if (has_type < 0)
 792                         die(_("cannot read existing object info %s"), oid_to_hex(oid));
 793                 if (has_type != type || has_size != size)
 794                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 795                 has_data = read_object_file(oid, &has_type, &has_size);
 796                 read_unlock();
 797                 if (!data)
 798                         data = new_data = get_data_from_pack(obj_entry);
 799                 if (!has_data)
 800                         die(_("cannot read existing object %s"), oid_to_hex(oid));
 801                 if (size != has_size || type != has_type ||
 802                     memcmp(data, has_data, size) != 0)
 803                         die(_("SHA1 COLLISION FOUND WITH %s !"), oid_to_hex(oid));
 804                 free(has_data);
 805         }
 806
 807         if (strict || do_fsck_object) {
 808                 read_lock();
 809                 if (type == OBJ_BLOB) {
 810                         struct blob *blob = lookup_blob(the_repository, oid);
 811                         if (blob)
 812                                 blob->object.flags |= FLAG_CHECKED;
 813                         else
 814                                 die(_("invalid blob object %s"), oid_to_hex(oid));
 815                         if (do_fsck_object &&
 816                             fsck_object(&blob->object, (void *)data, size, &fsck_options))
 817                                 die(_("fsck error in packed object"));
 818                 } else {
 819                         struct object *obj;
 820                         int eaten;
 821                         void *buf = (void *) data;
 822
 823                         assert(data && "data can only be NULL for large _blobs_");
 824
 825                         /*
 826                          * we do not need to free the memory here, as the
 827                          * buf is deleted by the caller.
 828                          */
 829                         obj = parse_object_buffer(the_repository, oid, type,
 830                                                   size, buf,
 831                                                   &eaten);
 832                         if (!obj)
 833                                 die(_("invalid %s"), type_name(type));
 834                         if (do_fsck_object &&
 835                             fsck_object(obj, buf, size, &fsck_options))
 836                                 die(_("fsck error in packed object"));
 837                         if (strict && fsck_walk(obj, NULL, &fsck_options))
 838                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 839
 840                         if (obj->type == OBJ_TREE) {
 841                                 struct tree *item = (struct tree *) obj;
 842                                 item->buffer = NULL;
 843                                 obj->parsed = 0;
 844                         }
 845                         if (obj->type == OBJ_COMMIT) {
 846                                 struct commit *commit = (struct commit *) obj;
 847                                 if (detach_commit_buffer(commit, NULL) != data)
 848                                         BUG("parse_object_buffer transmogrified our buffer");
 849                         }
 850                         obj->flags |= FLAG_CHECKED;
 851                 }
 852                 read_unlock();
 853         }
 854
 855         free(new_data);
 856 }
 857
 858 /*
 859  * This function is part of find_unresolved_deltas(). There are two
 860  * walkers going in the opposite ways.
 861  *
 862  * The first one in find_unresolved_deltas() traverses down from
 863  * parent node to children, deflating nodes along the way. However,
 864  * memory for deflated nodes is limited by delta_base_cache_limit, so
 865  * at some point parent node's deflated content may be freed.
 866  *
 867  * The second walker is this function, which goes from current node up
 868  * to top parent if necessary to deflate the node. In normal
 869  * situation, its parent node would be already deflated, so it just
 870  * needs to apply delta.
 871  *
 872  * In the worst case scenario, parent node is no longer deflated because
 873  * we're running out of delta_base_cache_limit; we need to re-deflate
 874  * parents, possibly up to the top base.
 875  *
 876  * All deflated objects here are subject to be freed if we exceed
 877  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 878  * just need to make sure the last node is not freed.
 879  */
 880 static void *get_base_data(struct base_data *c)
 881 {
 882         if (!c->data) {
 883                 struct object_entry *obj = c->obj;
 884                 struct base_data **delta = NULL;
 885                 int delta_nr = 0, delta_alloc = 0;
 886
 887                 while (is_delta_type(c->obj->type) && !c->data) {
 888                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 889                         delta[delta_nr++] = c;
 890                         c = c->base;
 891                 }
 892                 if (!delta_nr) {
 893                         c->data = get_data_from_pack(obj);
 894                         c->size = obj->size;
 895                         get_thread_data()->base_cache_used += c->size;
 896                         prune_base_data(c);
 897                 }
 898                 for (; delta_nr > 0; delta_nr--) {
 899                         void *base, *raw;
 900                         c = delta[delta_nr - 1];
 901                         obj = c->obj;
 902                         base = get_base_data(c->base);
 903                         raw = get_data_from_pack(obj);
 904                         c->data = patch_delta(
 905                                 base, c->base->size,
 906                                 raw, obj->size,
 907                                 &c->size);
 908                         free(raw);
 909                         if (!c->data)
 910                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 911                         get_thread_data()->base_cache_used += c->size;
 912                         prune_base_data(c);
 913                 }
 914                 free(delta);
 915         }
 916         return c->data;
 917 }
 918
 919 static void resolve_delta(struct object_entry *delta_obj,
 920                           struct base_data *base, struct base_data *result)
 921 {
 922         void *base_data, *delta_data;
 923
 924         if (show_stat) {
 925                 int i = delta_obj - objects;
 926                 int j = base->obj - objects;
 927                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 928                 deepest_delta_lock();
 929                 if (deepest_delta < obj_stat[i].delta_depth)
 930                         deepest_delta = obj_stat[i].delta_depth;
 931                 deepest_delta_unlock();
 932                 obj_stat[i].base_object_no = j;
 933         }
 934         delta_data = get_data_from_pack(delta_obj);
 935         base_data = get_base_data(base);
 936         result->obj = delta_obj;
 937         result->data = patch_delta(base_data, base->size,
 938                                    delta_data, delta_obj->size, &result->size);
 939         free(delta_data);
 940         if (!result->data)
 941                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 942         hash_object_file(result->data, result->size,
 943                          type_name(delta_obj->real_type), &delta_obj->idx.oid);
 944         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 945                     &delta_obj->idx.oid);
 946         counter_lock();
 947         nr_resolved_deltas++;
 948         counter_unlock();
 949 }
 950
 951 /*
 952  * Standard boolean compare-and-swap: atomically check whether "*type" is
 953  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 954  * and return false.
 955  */
 956 static int compare_and_swap_type(signed char *type,
 957                                  enum object_type want,
 958                                  enum object_type set)
 959 {
 960         enum object_type old;
 961
 962         type_cas_lock();
 963         old = *type;
 964         if (old == want)
 965                 *type = set;
 966         type_cas_unlock();
 967
 968         return old == want;
 969 }
 970
 971 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 972                                                   struct base_data *prev_base)
 973 {
 974         if (base->ref_last == -1 && base->ofs_last == -1) {
 975                 find_ref_delta_children(&base->obj->idx.oid,
 976                                         &base->ref_first, &base->ref_last,
 977                                         OBJ_REF_DELTA);
 978
 979                 find_ofs_delta_children(base->obj->idx.offset,
 980                                         &base->ofs_first, &base->ofs_last,
 981                                         OBJ_OFS_DELTA);
 982
 983                 if (base->ref_last == -1 && base->ofs_last == -1) {
 984                         free(base->data);
 985                         return NULL;
 986                 }
 987
 988                 link_base_data(prev_base, base);
 989         }
 990
 991         if (base->ref_first <= base->ref_last) {
 992                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
 993                 struct base_data *result = alloc_base_data();
 994
 995                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
 996                                            base->obj->real_type))
 997                         BUG("child->real_type != OBJ_REF_DELTA");
 998
 999                 resolve_delta(child, base, result);
1000                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1001                         free_base_data(base);
1002
1003                 base->ref_first++;
1004                 return result;
1005         }
1006
1007         if (base->ofs_first <= base->ofs_last) {
1008                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1009                 struct base_data *result = alloc_base_data();
1010
1011                 assert(child->real_type == OBJ_OFS_DELTA);
1012                 child->real_type = base->obj->real_type;
1013                 resolve_delta(child, base, result);
1014                 if (base->ofs_first == base->ofs_last)
1015                         free_base_data(base);
1016
1017                 base->ofs_first++;
1018                 return result;
1019         }
1020
1021         unlink_base_data(base);
1022         return NULL;
1023 }
1024
1025 static void find_unresolved_deltas(struct base_data *base)
1026 {
1027         struct base_data *new_base, *prev_base = NULL;
1028         for (;;) {
1029                 new_base = find_unresolved_deltas_1(base, prev_base);
1030
1031                 if (new_base) {
1032                         prev_base = base;
1033                         base = new_base;
1034                 } else {
1035                         free(base);
1036                         base = prev_base;
1037                         if (!base)
1038                                 return;
1039                         prev_base = base->base;
1040                 }
1041         }
1042 }
1043
1044 static int compare_ofs_delta_entry(const void *a, const void *b)
1045 {
1046         const struct ofs_delta_entry *delta_a = a;
1047         const struct ofs_delta_entry *delta_b = b;
1048
1049         return delta_a->offset < delta_b->offset ? -1 :
1050                delta_a->offset > delta_b->offset ?  1 :
1051                0;
1052 }
1053
1054 static int compare_ref_delta_entry(const void *a, const void *b)
1055 {
1056         const struct ref_delta_entry *delta_a = a;
1057         const struct ref_delta_entry *delta_b = b;
1058
1059         return oidcmp(&delta_a->oid, &delta_b->oid);
1060 }
1061
1062 static void resolve_base(struct object_entry *obj)
1063 {
1064         struct base_data *base_obj = alloc_base_data();
1065         base_obj->obj = obj;
1066         base_obj->data = NULL;
1067         find_unresolved_deltas(base_obj);
1068 }
1069
1070 static void *threaded_second_pass(void *data)
1071 {
1072         set_thread_data(data);
1073         for (;;) {
1074                 int i;
1075                 counter_lock();
1076                 display_progress(progress, nr_resolved_deltas);
1077                 counter_unlock();
1078                 work_lock();
1079                 while (nr_dispatched < nr_objects &&
1080                        is_delta_type(objects[nr_dispatched].type))
1081                         nr_dispatched++;
1082                 if (nr_dispatched >= nr_objects) {
1083                         work_unlock();
1084                         break;
1085                 }
1086                 i = nr_dispatched++;
1087                 work_unlock();
1088
1089                 resolve_base(&objects[i]);
1090         }
1091         return NULL;
1092 }
1093
1094 /*
1095  * First pass:
1096  * - find locations of all objects;
1097  * - calculate SHA1 of all non-delta objects;
1098  * - remember base (SHA1 or offset) for all deltas.
1099  */
1100 static void parse_pack_objects(unsigned char *hash)
1101 {
1102         int i, nr_delays = 0;
1103         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1104         struct object_id ref_delta_oid;
1105         struct stat st;
1106
1107         if (verbose)
1108                 progress = start_progress(
1109                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1110                                 nr_objects);
1111         for (i = 0; i < nr_objects; i++) {
1112                 struct object_entry *obj = &objects[i];
1113                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1114                                               &ref_delta_oid,
1115                                               &obj->idx.oid);
1116                 obj->real_type = obj->type;
1117                 if (obj->type == OBJ_OFS_DELTA) {
1118                         nr_ofs_deltas++;
1119                         ofs_delta->obj_no = i;
1120                         ofs_delta++;
1121                 } else if (obj->type == OBJ_REF_DELTA) {
1122                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1123                         oidcpy(&ref_deltas[nr_ref_deltas].oid, &ref_delta_oid);
1124                         ref_deltas[nr_ref_deltas].obj_no = i;
1125                         nr_ref_deltas++;
1126                 } else if (!data) {
1127                         /* large blobs, check later */
1128                         obj->real_type = OBJ_BAD;
1129                         nr_delays++;
1130                 } else
1131                         sha1_object(data, NULL, obj->size, obj->type,
1132                                     &obj->idx.oid);
1133                 free(data);
1134                 display_progress(progress, i+1);
1135         }
1136         objects[i].idx.offset = consumed_bytes;
1137         stop_progress(&progress);
1138
1139         /* Check pack integrity */
1140         flush();
1141         the_hash_algo->final_fn(hash, &input_ctx);
1142         if (!hasheq(fill(the_hash_algo->rawsz), hash))
1143                 die(_("pack is corrupted (SHA1 mismatch)"));
1144         use(the_hash_algo->rawsz);
1145
1146         /* If input_fd is a file, we should have reached its end now. */
1147         if (fstat(input_fd, &st))
1148                 die_errno(_("cannot fstat packfile"));
1149         if (S_ISREG(st.st_mode) &&
1150                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1151                 die(_("pack has junk at the end"));
1152
1153         for (i = 0; i < nr_objects; i++) {
1154                 struct object_entry *obj = &objects[i];
1155                 if (obj->real_type != OBJ_BAD)
1156                         continue;
1157                 obj->real_type = obj->type;
1158                 sha1_object(NULL, obj, obj->size, obj->type,
1159                             &obj->idx.oid);
1160                 nr_delays--;
1161         }
1162         if (nr_delays)
1163                 die(_("confusion beyond insanity in parse_pack_objects()"));
1164 }
1165
1166 /*
1167  * Second pass:
1168  * - for all non-delta objects, look if it is used as a base for
1169  *   deltas;
1170  * - if used as a base, uncompress the object and apply all deltas,
1171  *   recursively checking if the resulting object is used as a base
1172  *   for some more deltas.
1173  */
1174 static void resolve_deltas(void)
1175 {
1176         int i;
1177
1178         if (!nr_ofs_deltas && !nr_ref_deltas)
1179                 return;
1180
1181         /* Sort deltas by base SHA1/offset for fast searching */
1182         QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1183         QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1184
1185         if (verbose || show_resolving_progress)
1186                 progress = start_progress(_("Resolving deltas"),
1187                                           nr_ref_deltas + nr_ofs_deltas);
1188
1189         nr_dispatched = 0;
1190         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1191                 init_thread();
1192                 for (i = 0; i < nr_threads; i++) {
1193                         int ret = pthread_create(&thread_data[i].thread, NULL,
1194                                                  threaded_second_pass, thread_data + i);
1195                         if (ret)
1196                                 die(_("unable to create thread: %s"),
1197                                     strerror(ret));
1198                 }
1199                 for (i = 0; i < nr_threads; i++)
1200                         pthread_join(thread_data[i].thread, NULL);
1201                 cleanup_thread();
1202                 return;
1203         }
1204
1205         for (i = 0; i < nr_objects; i++) {
1206                 struct object_entry *obj = &objects[i];
1207
1208                 if (is_delta_type(obj->type))
1209                         continue;
1210                 resolve_base(obj);
1211                 display_progress(progress, nr_resolved_deltas);
1212         }
1213 }
1214
1215 /*
1216  * Third pass:
1217  * - append objects to convert thin pack to full pack if required
1218  * - write the final pack hash
1219  */
1220 static void fix_unresolved_deltas(struct hashfile *f);
1221 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_hash)
1222 {
1223         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1224                 stop_progress(&progress);
1225                 /* Flush remaining pack final hash. */
1226                 flush();
1227                 return;
1228         }
1229
1230         if (fix_thin_pack) {
1231                 struct hashfile *f;
1232                 unsigned char read_hash[GIT_MAX_RAWSZ], tail_hash[GIT_MAX_RAWSZ];
1233                 struct strbuf msg = STRBUF_INIT;
1234                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1235                 int nr_objects_initial = nr_objects;
1236                 if (nr_unresolved <= 0)
1237                         die(_("confusion beyond insanity"));
1238                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1239                 memset(objects + nr_objects + 1, 0,
1240                        nr_unresolved * sizeof(*objects));
1241                 f = hashfd(output_fd, curr_pack);
1242                 fix_unresolved_deltas(f);
1243                 strbuf_addf(&msg, Q_("completed with %d local object",
1244                                      "completed with %d local objects",
1245                                      nr_objects - nr_objects_initial),
1246                             nr_objects - nr_objects_initial);
1247                 stop_progress_msg(&progress, msg.buf);
1248                 strbuf_release(&msg);
1249                 finalize_hashfile(f, tail_hash, 0);
1250                 hashcpy(read_hash, pack_hash);
1251                 fixup_pack_header_footer(output_fd, pack_hash,
1252                                          curr_pack, nr_objects,
1253                                          read_hash, consumed_bytes-the_hash_algo->rawsz);
1254                 if (!hasheq(read_hash, tail_hash))
1255                         die(_("Unexpected tail checksum for %s "
1256                               "(disk corruption?)"), curr_pack);
1257         }
1258         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1259                 die(Q_("pack has %d unresolved delta",
1260                        "pack has %d unresolved deltas",
1261                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1262                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1263 }
1264
1265 static int write_compressed(struct hashfile *f, void *in, unsigned int size)
1266 {
1267         git_zstream stream;
1268         int status;
1269         unsigned char outbuf[4096];
1270
1271         git_deflate_init(&stream, zlib_compression_level);
1272         stream.next_in = in;
1273         stream.avail_in = size;
1274
1275         do {
1276                 stream.next_out = outbuf;
1277                 stream.avail_out = sizeof(outbuf);
1278                 status = git_deflate(&stream, Z_FINISH);
1279                 hashwrite(f, outbuf, sizeof(outbuf) - stream.avail_out);
1280         } while (status == Z_OK);
1281
1282         if (status != Z_STREAM_END)
1283                 die(_("unable to deflate appended object (%d)"), status);
1284         size = stream.total_out;
1285         git_deflate_end(&stream);
1286         return size;
1287 }
1288
1289 static struct object_entry *append_obj_to_pack(struct hashfile *f,
1290                                const unsigned char *sha1, void *buf,
1291                                unsigned long size, enum object_type type)
1292 {
1293         struct object_entry *obj = &objects[nr_objects++];
1294         unsigned char header[10];
1295         unsigned long s = size;
1296         int n = 0;
1297         unsigned char c = (type << 4) | (s & 15);
1298         s >>= 4;
1299         while (s) {
1300                 header[n++] = c | 0x80;
1301                 c = s & 0x7f;
1302                 s >>= 7;
1303         }
1304         header[n++] = c;
1305         crc32_begin(f);
1306         hashwrite(f, header, n);
1307         obj[0].size = size;
1308         obj[0].hdr_size = n;
1309         obj[0].type = type;
1310         obj[0].real_type = type;
1311         obj[1].idx.offset = obj[0].idx.offset + n;
1312         obj[1].idx.offset += write_compressed(f, buf, size);
1313         obj[0].idx.crc32 = crc32_end(f);
1314         hashflush(f);
1315         hashcpy(obj->idx.oid.hash, sha1);
1316         return obj;
1317 }
1318
1319 static int delta_pos_compare(const void *_a, const void *_b)
1320 {
1321         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1322         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1323         return a->obj_no - b->obj_no;
1324 }
1325
1326 static void fix_unresolved_deltas(struct hashfile *f)
1327 {
1328         struct ref_delta_entry **sorted_by_pos;
1329         int i;
1330
1331         /*
1332          * Since many unresolved deltas may well be themselves base objects
1333          * for more unresolved deltas, we really want to include the
1334          * smallest number of base objects that would cover as much delta
1335          * as possible by picking the
1336          * trunc deltas first, allowing for other deltas to resolve without
1337          * additional base objects.  Since most base objects are to be found
1338          * before deltas depending on them, a good heuristic is to start
1339          * resolving deltas in the same order as their position in the pack.
1340          */
1341         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1342         for (i = 0; i < nr_ref_deltas; i++)
1343                 sorted_by_pos[i] = &ref_deltas[i];
1344         QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
1345
1346         for (i = 0; i < nr_ref_deltas; i++) {
1347                 struct ref_delta_entry *d = sorted_by_pos[i];
1348                 enum object_type type;
1349                 struct base_data *base_obj = alloc_base_data();
1350
1351                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1352                         continue;
1353                 base_obj->data = read_object_file(&d->oid, &type,
1354                                                   &base_obj->size);
1355                 if (!base_obj->data)
1356                         continue;
1357
1358                 if (check_object_signature(&d->oid, base_obj->data,
1359                                 base_obj->size, type_name(type)))
1360                         die(_("local object %s is corrupt"), oid_to_hex(&d->oid));
1361                 base_obj->obj = append_obj_to_pack(f, d->oid.hash,
1362                                         base_obj->data, base_obj->size, type);
1363                 find_unresolved_deltas(base_obj);
1364                 display_progress(progress, nr_resolved_deltas);
1365         }
1366         free(sorted_by_pos);
1367 }
1368
1369 static const char *derive_filename(const char *pack_name, const char *suffix,
1370                                    struct strbuf *buf)
1371 {
1372         size_t len;
1373         if (!strip_suffix(pack_name, ".pack", &len))
1374                 die(_("packfile name '%s' does not end with '.pack'"),
1375                     pack_name);
1376         strbuf_add(buf, pack_name, len);
1377         strbuf_addch(buf, '.');
1378         strbuf_addstr(buf, suffix);
1379         return buf->buf;
1380 }
1381
1382 static void write_special_file(const char *suffix, const char *msg,
1383                                const char *pack_name, const unsigned char *hash,
1384                                const char **report)
1385 {
1386         struct strbuf name_buf = STRBUF_INIT;
1387         const char *filename;
1388         int fd;
1389         int msg_len = strlen(msg);
1390
1391         if (pack_name)
1392                 filename = derive_filename(pack_name, suffix, &name_buf);
1393         else
1394                 filename = odb_pack_name(&name_buf, hash, suffix);
1395
1396         fd = odb_pack_keep(filename);
1397         if (fd < 0) {
1398                 if (errno != EEXIST)
1399                         die_errno(_("cannot write %s file '%s'"),
1400                                   suffix, filename);
1401         } else {
1402                 if (msg_len > 0) {
1403                         write_or_die(fd, msg, msg_len);
1404                         write_or_die(fd, "\n", 1);
1405                 }
1406                 if (close(fd) != 0)
1407                         die_errno(_("cannot close written %s file '%s'"),
1408                                   suffix, filename);
1409                 if (report)
1410                         *report = suffix;
1411         }
1412         strbuf_release(&name_buf);
1413 }
1414
1415 static void final(const char *final_pack_name, const char *curr_pack_name,
1416                   const char *final_index_name, const char *curr_index_name,
1417                   const char *keep_msg, const char *promisor_msg,
1418                   unsigned char *hash)
1419 {
1420         const char *report = "pack";
1421         struct strbuf pack_name = STRBUF_INIT;
1422         struct strbuf index_name = STRBUF_INIT;
1423         int err;
1424
1425         if (!from_stdin) {
1426                 close(input_fd);
1427         } else {
1428                 fsync_or_die(output_fd, curr_pack_name);
1429                 err = close(output_fd);
1430                 if (err)
1431                         die_errno(_("error while closing pack file"));
1432         }
1433
1434         if (keep_msg)
1435                 write_special_file("keep", keep_msg, final_pack_name, hash,
1436                                    &report);
1437         if (promisor_msg)
1438                 write_special_file("promisor", promisor_msg, final_pack_name,
1439                                    hash, NULL);
1440
1441         if (final_pack_name != curr_pack_name) {
1442                 if (!final_pack_name)
1443                         final_pack_name = odb_pack_name(&pack_name, hash, "pack");
1444                 if (finalize_object_file(curr_pack_name, final_pack_name))
1445                         die(_("cannot store pack file"));
1446         } else if (from_stdin)
1447                 chmod(final_pack_name, 0444);
1448
1449         if (final_index_name != curr_index_name) {
1450                 if (!final_index_name)
1451                         final_index_name = odb_pack_name(&index_name, hash, "idx");
1452                 if (finalize_object_file(curr_index_name, final_index_name))
1453                         die(_("cannot store index file"));
1454         } else
1455                 chmod(final_index_name, 0444);
1456
1457         if (do_fsck_object) {
1458                 struct packed_git *p;
1459                 p = add_packed_git(final_index_name, strlen(final_index_name), 0);
1460                 if (p)
1461                         install_packed_git(the_repository, p);
1462         }
1463
1464         if (!from_stdin) {
1465                 printf("%s\n", sha1_to_hex(hash));
1466         } else {
1467                 struct strbuf buf = STRBUF_INIT;
1468
1469                 strbuf_addf(&buf, "%s\t%s\n", report, sha1_to_hex(hash));
1470                 write_or_die(1, buf.buf, buf.len);
1471                 strbuf_release(&buf);
1472
1473                 /*
1474                  * Let's just mimic git-unpack-objects here and write
1475                  * the last part of the input buffer to stdout.
1476                  */
1477                 while (input_len) {
1478                         err = xwrite(1, input_buffer + input_offset, input_len);
1479                         if (err <= 0)
1480                                 break;
1481                         input_len -= err;
1482                         input_offset += err;
1483                 }
1484         }
1485
1486         strbuf_release(&index_name);
1487         strbuf_release(&pack_name);
1488 }
1489
1490 static int git_index_pack_config(const char *k, const char *v, void *cb)
1491 {
1492         struct pack_idx_option *opts = cb;
1493
1494         if (!strcmp(k, "pack.indexversion")) {
1495                 opts->version = git_config_int(k, v);
1496                 if (opts->version > 2)
1497                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1498                 return 0;
1499         }
1500         if (!strcmp(k, "pack.threads")) {
1501                 nr_threads = git_config_int(k, v);
1502                 if (nr_threads < 0)
1503                         die(_("invalid number of threads specified (%d)"),
1504                             nr_threads);
1505                 if (!HAVE_THREADS && nr_threads != 1) {
1506                         warning(_("no threads support, ignoring %s"), k);
1507                         nr_threads = 1;
1508                 }
1509                 return 0;
1510         }
1511         return git_default_config(k, v, cb);
1512 }
1513
1514 static int cmp_uint32(const void *a_, const void *b_)
1515 {
1516         uint32_t a = *((uint32_t *)a_);
1517         uint32_t b = *((uint32_t *)b_);
1518
1519         return (a < b) ? -1 : (a != b);
1520 }
1521
1522 static void read_v2_anomalous_offsets(struct packed_git *p,
1523                                       struct pack_idx_option *opts)
1524 {
1525         const uint32_t *idx1, *idx2;
1526         uint32_t i;
1527         const uint32_t hashwords = the_hash_algo->rawsz / sizeof(uint32_t);
1528
1529         /* The address of the 4-byte offset table */
1530         idx1 = (((const uint32_t *)p->index_data)
1531                 + 2 /* 8-byte header */
1532                 + 256 /* fan out */
1533                 + hashwords * p->num_objects /* object ID table */
1534                 + p->num_objects /* CRC32 table */
1535                 );
1536
1537         /* The address of the 8-byte offset table */
1538         idx2 = idx1 + p->num_objects;
1539
1540         for (i = 0; i < p->num_objects; i++) {
1541                 uint32_t off = ntohl(idx1[i]);
1542                 if (!(off & 0x80000000))
1543                         continue;
1544                 off = off & 0x7fffffff;
1545                 check_pack_index_ptr(p, &idx2[off * 2]);
1546                 if (idx2[off * 2])
1547                         continue;
1548                 /*
1549                  * The real offset is ntohl(idx2[off * 2]) in high 4
1550                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1551                  * octets.  But idx2[off * 2] is Zero!!!
1552                  */
1553                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1554                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1555         }
1556
1557         QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32);
1558 }
1559
1560 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1561 {
1562         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1563
1564         if (!p)
1565                 die(_("Cannot open existing pack file '%s'"), pack_name);
1566         if (open_pack_index(p))
1567                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1568
1569         /* Read the attributes from the existing idx file */
1570         opts->version = p->index_version;
1571
1572         if (opts->version == 2)
1573                 read_v2_anomalous_offsets(p, opts);
1574
1575         /*
1576          * Get rid of the idx file as we do not need it anymore.
1577          * NEEDSWORK: extract this bit from free_pack_by_name() in
1578          * sha1-file.c, perhaps?  It shouldn't matter very much as we
1579          * know we haven't installed this pack (hence we never have
1580          * read anything from it).
1581          */
1582         close_pack_index(p);
1583         free(p);
1584 }
1585
1586 static void show_pack_info(int stat_only)
1587 {
1588         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1589         unsigned long *chain_histogram = NULL;
1590
1591         if (deepest_delta)
1592                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1593
1594         for (i = 0; i < nr_objects; i++) {
1595                 struct object_entry *obj = &objects[i];
1596
1597                 if (is_delta_type(obj->type))
1598                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1599                 if (stat_only)
1600                         continue;
1601                 printf("%s %-6s %"PRIuMAX" %"PRIuMAX" %"PRIuMAX,
1602                        oid_to_hex(&obj->idx.oid),
1603                        type_name(obj->real_type), (uintmax_t)obj->size,
1604                        (uintmax_t)(obj[1].idx.offset - obj->idx.offset),
1605                        (uintmax_t)obj->idx.offset);
1606                 if (is_delta_type(obj->type)) {
1607                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1608                         printf(" %u %s", obj_stat[i].delta_depth,
1609                                oid_to_hex(&bobj->idx.oid));
1610                 }
1611                 putchar('\n');
1612         }
1613
1614         if (baseobjects)
1615                 printf_ln(Q_("non delta: %d object",
1616                              "non delta: %d objects",
1617                              baseobjects),
1618                           baseobjects);
1619         for (i = 0; i < deepest_delta; i++) {
1620                 if (!chain_histogram[i])
1621                         continue;
1622                 printf_ln(Q_("chain length = %d: %lu object",
1623                              "chain length = %d: %lu objects",
1624                              chain_histogram[i]),
1625                           i + 1,
1626                           chain_histogram[i]);
1627         }
1628 }
1629
1630 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1631 {
1632         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1633         const char *curr_index;
1634         const char *index_name = NULL, *pack_name = NULL;
1635         const char *keep_msg = NULL;
1636         const char *promisor_msg = NULL;
1637         struct strbuf index_name_buf = STRBUF_INIT;
1638         struct pack_idx_entry **idx_objects;
1639         struct pack_idx_option opts;
1640         unsigned char pack_hash[GIT_MAX_RAWSZ];
1641         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1642         int report_end_of_input = 0;
1643
1644         /*
1645          * index-pack never needs to fetch missing objects, since it only
1646          * accesses the repo to do hash collision checks
1647          */
1648         fetch_if_missing = 0;
1649
1650         if (argc == 2 && !strcmp(argv[1], "-h"))
1651                 usage(index_pack_usage);
1652
1653         read_replace_refs = 0;
1654         fsck_options.walk = mark_link;
1655
1656         reset_pack_idx_option(&opts);
1657         git_config(git_index_pack_config, &opts);
1658         if (prefix && chdir(prefix))
1659                 die(_("Cannot come back to cwd"));
1660
1661         for (i = 1; i < argc; i++) {
1662                 const char *arg = argv[i];
1663
1664                 if (*arg == '-') {
1665                         if (!strcmp(arg, "--stdin")) {
1666                                 from_stdin = 1;
1667                         } else if (!strcmp(arg, "--fix-thin")) {
1668                                 fix_thin_pack = 1;
1669                         } else if (skip_to_optional_arg(arg, "--strict", &arg)) {
1670                                 strict = 1;
1671                                 do_fsck_object = 1;
1672                                 fsck_set_msg_types(&fsck_options, arg);
1673                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1674                                 strict = 1;
1675                                 check_self_contained_and_connected = 1;
1676                         } else if (!strcmp(arg, "--fsck-objects")) {
1677                                 do_fsck_object = 1;
1678                         } else if (!strcmp(arg, "--verify")) {
1679                                 verify = 1;
1680                         } else if (!strcmp(arg, "--verify-stat")) {
1681                                 verify = 1;
1682                                 show_stat = 1;
1683                         } else if (!strcmp(arg, "--verify-stat-only")) {
1684                                 verify = 1;
1685                                 show_stat = 1;
1686                                 stat_only = 1;
1687                         } else if (skip_to_optional_arg(arg, "--keep", &keep_msg)) {
1688                                 ; /* nothing to do */
1689                         } else if (skip_to_optional_arg(arg, "--promisor", &promisor_msg)) {
1690                                 ; /* already parsed */
1691                         } else if (starts_with(arg, "--threads=")) {
1692                                 char *end;
1693                                 nr_threads = strtoul(arg+10, &end, 0);
1694                                 if (!arg[10] || *end || nr_threads < 0)
1695                                         usage(index_pack_usage);
1696                                 if (!HAVE_THREADS && nr_threads != 1) {
1697                                         warning(_("no threads support, ignoring %s"), arg);
1698                                         nr_threads = 1;
1699                                 }
1700                         } else if (starts_with(arg, "--pack_header=")) {
1701                                 struct pack_header *hdr;
1702                                 char *c;
1703
1704                                 hdr = (struct pack_header *)input_buffer;
1705                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1706                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1707                                 if (*c != ',')
1708                                         die(_("bad %s"), arg);
1709                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1710                                 if (*c)
1711                                         die(_("bad %s"), arg);
1712                                 input_len = sizeof(*hdr);
1713                         } else if (!strcmp(arg, "-v")) {
1714                                 verbose = 1;
1715                         } else if (!strcmp(arg, "--show-resolving-progress")) {
1716                                 show_resolving_progress = 1;
1717                         } else if (!strcmp(arg, "--report-end-of-input")) {
1718                                 report_end_of_input = 1;
1719                         } else if (!strcmp(arg, "-o")) {
1720                                 if (index_name || (i+1) >= argc)
1721                                         usage(index_pack_usage);
1722                                 index_name = argv[++i];
1723                         } else if (starts_with(arg, "--index-version=")) {
1724                                 char *c;
1725                                 opts.version = strtoul(arg + 16, &c, 10);
1726                                 if (opts.version > 2)
1727                                         die(_("bad %s"), arg);
1728                                 if (*c == ',')
1729                                         opts.off32_limit = strtoul(c+1, &c, 0);
1730                                 if (*c || opts.off32_limit & 0x80000000)
1731                                         die(_("bad %s"), arg);
1732                         } else if (skip_prefix(arg, "--max-input-size=", &arg)) {
1733                                 max_input_size = strtoumax(arg, NULL, 10);
1734                         } else
1735                                 usage(index_pack_usage);
1736                         continue;
1737                 }
1738
1739                 if (pack_name)
1740                         usage(index_pack_usage);
1741                 pack_name = arg;
1742         }
1743
1744         if (!pack_name && !from_stdin)
1745                 usage(index_pack_usage);
1746         if (fix_thin_pack && !from_stdin)
1747                 die(_("--fix-thin cannot be used without --stdin"));
1748         if (from_stdin && !startup_info->have_repository)
1749                 die(_("--stdin requires a git repository"));
1750         if (!index_name && pack_name)
1751                 index_name = derive_filename(pack_name, "idx", &index_name_buf);
1752
1753         if (verify) {
1754                 if (!index_name)
1755                         die(_("--verify with no packfile name given"));
1756                 read_idx_option(&opts, index_name);
1757                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1758         }
1759         if (strict)
1760                 opts.flags |= WRITE_IDX_STRICT;
1761
1762         if (HAVE_THREADS && !nr_threads) {
1763                 nr_threads = online_cpus();
1764                 /* An experiment showed that more threads does not mean faster */
1765                 if (nr_threads > 3)
1766                         nr_threads = 3;
1767         }
1768
1769         curr_pack = open_pack_file(pack_name);
1770         parse_pack_header();
1771         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1772         if (show_stat)
1773                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1774         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1775         parse_pack_objects(pack_hash);
1776         if (report_end_of_input)
1777                 write_in_full(2, "\0", 1);
1778         resolve_deltas();
1779         conclude_pack(fix_thin_pack, curr_pack, pack_hash);
1780         free(ofs_deltas);
1781         free(ref_deltas);
1782         if (strict)
1783                 foreign_nr = check_objects();
1784
1785         if (show_stat)
1786                 show_pack_info(stat_only);
1787
1788         ALLOC_ARRAY(idx_objects, nr_objects);
1789         for (i = 0; i < nr_objects; i++)
1790                 idx_objects[i] = &objects[i].idx;
1791         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_hash);
1792         free(idx_objects);
1793
1794         if (!verify)
1795                 final(pack_name, curr_pack,
1796                       index_name, curr_index,
1797                       keep_msg, promisor_msg,
1798                       pack_hash);
1799         else
1800                 close(input_fd);
1801
1802         if (do_fsck_object && fsck_finish(&fsck_options))
1803                 die(_("fsck error in pack objects"));
1804
1805         free(objects);
1806         strbuf_release(&index_name_buf);
1807         if (pack_name == NULL)
1808                 free((void *) curr_pack);
1809         if (index_name == NULL)
1810                 free((void *) curr_index);
1811
1812         /*
1813          * Let the caller know this pack is not self contained
1814          */
1815         if (check_self_contained_and_connected && foreign_nr)
1816                 return 1;
1817
1818         return 0;
1819 }