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

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12 #include "streaming.h"
  13 #include "thread-utils.h"
  14
  15 static const char index_pack_usage[] =
  16 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  17
  18 struct object_entry {
  19         struct pack_idx_entry idx;
  20         unsigned long size;
  21         unsigned char hdr_size;
  22         signed char type;
  23         signed char real_type;
  24 };
  25
  26 struct object_stat {
  27         unsigned delta_depth;
  28         int base_object_no;
  29 };
  30
  31 struct base_data {
  32         struct base_data *base;
  33         struct base_data *child;
  34         struct object_entry *obj;
  35         void *data;
  36         unsigned long size;
  37         int ref_first, ref_last;
  38         int ofs_first, ofs_last;
  39 };
  40
  41 struct thread_local {
  42 #ifndef NO_PTHREADS
  43         pthread_t thread;
  44 #endif
  45         struct base_data *base_cache;
  46         size_t base_cache_used;
  47         int pack_fd;
  48 };
  49
  50 #define FLAG_LINK (1u<<20)
  51 #define FLAG_CHECKED (1u<<21)
  52
  53 struct ofs_delta_entry {
  54         off_t offset;
  55         int obj_no;
  56 };
  57
  58 struct ref_delta_entry {
  59         unsigned char sha1[20];
  60         int obj_no;
  61 };
  62
  63 static struct object_entry *objects;
  64 static struct object_stat *obj_stat;
  65 static struct ofs_delta_entry *ofs_deltas;
  66 static struct ref_delta_entry *ref_deltas;
  67 static struct thread_local nothread_data;
  68 static int nr_objects;
  69 static int nr_ofs_deltas;
  70 static int nr_ref_deltas;
  71 static int ref_deltas_alloc;
  72 static int nr_resolved_deltas;
  73 static int nr_threads;
  74
  75 static int from_stdin;
  76 static int strict;
  77 static int do_fsck_object;
  78 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  79 static int verbose;
  80 static int show_resolving_progress;
  81 static int show_stat;
  82 static int check_self_contained_and_connected;
  83
  84 static struct progress *progress;
  85
  86 /* We always read in 4kB chunks. */
  87 static unsigned char input_buffer[4096];
  88 static unsigned int input_offset, input_len;
  89 static off_t consumed_bytes;
  90 static off_t max_input_size;
  91 static unsigned deepest_delta;
  92 static git_SHA_CTX input_ctx;
  93 static uint32_t input_crc32;
  94 static int input_fd, output_fd;
  95 static const char *curr_pack;
  96
  97 #ifndef NO_PTHREADS
  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 #else
 179
 180 #define read_lock()
 181 #define read_unlock()
 182
 183 #define counter_lock()
 184 #define counter_unlock()
 185
 186 #define work_lock()
 187 #define work_unlock()
 188
 189 #define deepest_delta_lock()
 190 #define deepest_delta_unlock()
 191
 192 #define type_cas_lock()
 193 #define type_cas_unlock()
 194
 195 #endif
 196
 197
 198 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 199 {
 200         if (!obj)
 201                 return -1;
 202
 203         if (type != OBJ_ANY && obj->type != type)
 204                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 205
 206         obj->flags |= FLAG_LINK;
 207         return 0;
 208 }
 209
 210 /* The content of each linked object must have been checked
 211    or it must be already present in the object database */
 212 static unsigned check_object(struct object *obj)
 213 {
 214         if (!obj)
 215                 return 0;
 216
 217         if (!(obj->flags & FLAG_LINK))
 218                 return 0;
 219
 220         if (!(obj->flags & FLAG_CHECKED)) {
 221                 unsigned long size;
 222                 int type = sha1_object_info(obj->oid.hash, &size);
 223                 if (type <= 0)
 224                         die(_("did not receive expected object %s"),
 225                               oid_to_hex(&obj->oid));
 226                 if (type != obj->type)
 227                         die(_("object %s: expected type %s, found %s"),
 228                             oid_to_hex(&obj->oid),
 229                             typename(obj->type), typename(type));
 230                 obj->flags |= FLAG_CHECKED;
 231                 return 1;
 232         }
 233
 234         return 0;
 235 }
 236
 237 static unsigned check_objects(void)
 238 {
 239         unsigned i, max, foreign_nr = 0;
 240
 241         max = get_max_object_index();
 242         for (i = 0; i < max; i++)
 243                 foreign_nr += check_object(get_indexed_object(i));
 244         return foreign_nr;
 245 }
 246
 247
 248 /* Discard current buffer used content. */
 249 static void flush(void)
 250 {
 251         if (input_offset) {
 252                 if (output_fd >= 0)
 253                         write_or_die(output_fd, input_buffer, input_offset);
 254                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 255                 memmove(input_buffer, input_buffer + input_offset, input_len);
 256                 input_offset = 0;
 257         }
 258 }
 259
 260 /*
 261  * Make sure at least "min" bytes are available in the buffer, and
 262  * return the pointer to the buffer.
 263  */
 264 static void *fill(int min)
 265 {
 266         if (min <= input_len)
 267                 return input_buffer + input_offset;
 268         if (min > sizeof(input_buffer))
 269                 die(Q_("cannot fill %d byte",
 270                        "cannot fill %d bytes",
 271                        min),
 272                     min);
 273         flush();
 274         do {
 275                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 276                                 sizeof(input_buffer) - input_len);
 277                 if (ret <= 0) {
 278                         if (!ret)
 279                                 die(_("early EOF"));
 280                         die_errno(_("read error on input"));
 281                 }
 282                 input_len += ret;
 283                 if (from_stdin)
 284                         display_throughput(progress, consumed_bytes + input_len);
 285         } while (input_len < min);
 286         return input_buffer;
 287 }
 288
 289 static void use(int bytes)
 290 {
 291         if (bytes > input_len)
 292                 die(_("used more bytes than were available"));
 293         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 294         input_len -= bytes;
 295         input_offset += bytes;
 296
 297         /* make sure off_t is sufficiently large not to wrap */
 298         if (signed_add_overflows(consumed_bytes, bytes))
 299                 die(_("pack too large for current definition of off_t"));
 300         consumed_bytes += bytes;
 301         if (max_input_size && consumed_bytes > max_input_size)
 302                 die(_("pack exceeds maximum allowed size"));
 303 }
 304
 305 static const char *open_pack_file(const char *pack_name)
 306 {
 307         if (from_stdin) {
 308                 input_fd = 0;
 309                 if (!pack_name) {
 310                         static char tmp_file[PATH_MAX];
 311                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 312                                                 "pack/tmp_pack_XXXXXX");
 313                         pack_name = xstrdup(tmp_file);
 314                 } else
 315                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 316                 if (output_fd < 0)
 317                         die_errno(_("unable to create '%s'"), pack_name);
 318                 nothread_data.pack_fd = output_fd;
 319         } else {
 320                 input_fd = open(pack_name, O_RDONLY);
 321                 if (input_fd < 0)
 322                         die_errno(_("cannot open packfile '%s'"), pack_name);
 323                 output_fd = -1;
 324                 nothread_data.pack_fd = input_fd;
 325         }
 326         git_SHA1_Init(&input_ctx);
 327         return pack_name;
 328 }
 329
 330 static void parse_pack_header(void)
 331 {
 332         struct pack_header *hdr = fill(sizeof(struct pack_header));
 333
 334         /* Header consistency check */
 335         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 336                 die(_("pack signature mismatch"));
 337         if (!pack_version_ok(hdr->hdr_version))
 338                 die(_("pack version %"PRIu32" unsupported"),
 339                         ntohl(hdr->hdr_version));
 340
 341         nr_objects = ntohl(hdr->hdr_entries);
 342         use(sizeof(struct pack_header));
 343 }
 344
 345 static NORETURN void bad_object(off_t offset, const char *format,
 346                        ...) __attribute__((format (printf, 2, 3)));
 347
 348 static NORETURN void bad_object(off_t offset, const char *format, ...)
 349 {
 350         va_list params;
 351         char buf[1024];
 352
 353         va_start(params, format);
 354         vsnprintf(buf, sizeof(buf), format, params);
 355         va_end(params);
 356         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 357             (uintmax_t)offset, buf);
 358 }
 359
 360 static inline struct thread_local *get_thread_data(void)
 361 {
 362 #ifndef NO_PTHREADS
 363         if (threads_active)
 364                 return pthread_getspecific(key);
 365         assert(!threads_active &&
 366                "This should only be reached when all threads are gone");
 367 #endif
 368         return &nothread_data;
 369 }
 370
 371 #ifndef NO_PTHREADS
 372 static void set_thread_data(struct thread_local *data)
 373 {
 374         if (threads_active)
 375                 pthread_setspecific(key, data);
 376 }
 377 #endif
 378
 379 static struct base_data *alloc_base_data(void)
 380 {
 381         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 382         base->ref_last = -1;
 383         base->ofs_last = -1;
 384         return base;
 385 }
 386
 387 static void free_base_data(struct base_data *c)
 388 {
 389         if (c->data) {
 390                 free(c->data);
 391                 c->data = NULL;
 392                 get_thread_data()->base_cache_used -= c->size;
 393         }
 394 }
 395
 396 static void prune_base_data(struct base_data *retain)
 397 {
 398         struct base_data *b;
 399         struct thread_local *data = get_thread_data();
 400         for (b = data->base_cache;
 401              data->base_cache_used > delta_base_cache_limit && b;
 402              b = b->child) {
 403                 if (b->data && b != retain)
 404                         free_base_data(b);
 405         }
 406 }
 407
 408 static void link_base_data(struct base_data *base, struct base_data *c)
 409 {
 410         if (base)
 411                 base->child = c;
 412         else
 413                 get_thread_data()->base_cache = c;
 414
 415         c->base = base;
 416         c->child = NULL;
 417         if (c->data)
 418                 get_thread_data()->base_cache_used += c->size;
 419         prune_base_data(c);
 420 }
 421
 422 static void unlink_base_data(struct base_data *c)
 423 {
 424         struct base_data *base = c->base;
 425         if (base)
 426                 base->child = NULL;
 427         else
 428                 get_thread_data()->base_cache = NULL;
 429         free_base_data(c);
 430 }
 431
 432 static int is_delta_type(enum object_type type)
 433 {
 434         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 435 }
 436
 437 static void *unpack_entry_data(off_t offset, unsigned long size,
 438                                enum object_type type, unsigned char *sha1)
 439 {
 440         static char fixed_buf[8192];
 441         int status;
 442         git_zstream stream;
 443         void *buf;
 444         git_SHA_CTX c;
 445         char hdr[32];
 446         int hdrlen;
 447
 448         if (!is_delta_type(type)) {
 449                 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %lu", typename(type), size) + 1;
 450                 git_SHA1_Init(&c);
 451                 git_SHA1_Update(&c, hdr, hdrlen);
 452         } else
 453                 sha1 = NULL;
 454         if (type == OBJ_BLOB && size > big_file_threshold)
 455                 buf = fixed_buf;
 456         else
 457                 buf = xmallocz(size);
 458
 459         memset(&stream, 0, sizeof(stream));
 460         git_inflate_init(&stream);
 461         stream.next_out = buf;
 462         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 463
 464         do {
 465                 unsigned char *last_out = stream.next_out;
 466                 stream.next_in = fill(1);
 467                 stream.avail_in = input_len;
 468                 status = git_inflate(&stream, 0);
 469                 use(input_len - stream.avail_in);
 470                 if (sha1)
 471                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 472                 if (buf == fixed_buf) {
 473                         stream.next_out = buf;
 474                         stream.avail_out = sizeof(fixed_buf);
 475                 }
 476         } while (status == Z_OK);
 477         if (stream.total_out != size || status != Z_STREAM_END)
 478                 bad_object(offset, _("inflate returned %d"), status);
 479         git_inflate_end(&stream);
 480         if (sha1)
 481                 git_SHA1_Final(sha1, &c);
 482         return buf == fixed_buf ? NULL : buf;
 483 }
 484
 485 static void *unpack_raw_entry(struct object_entry *obj,
 486                               off_t *ofs_offset,
 487                               unsigned char *ref_sha1,
 488                               unsigned char *sha1)
 489 {
 490         unsigned char *p;
 491         unsigned long size, c;
 492         off_t base_offset;
 493         unsigned shift;
 494         void *data;
 495
 496         obj->idx.offset = consumed_bytes;
 497         input_crc32 = crc32(0, NULL, 0);
 498
 499         p = fill(1);
 500         c = *p;
 501         use(1);
 502         obj->type = (c >> 4) & 7;
 503         size = (c & 15);
 504         shift = 4;
 505         while (c & 0x80) {
 506                 p = fill(1);
 507                 c = *p;
 508                 use(1);
 509                 size += (c & 0x7f) << shift;
 510                 shift += 7;
 511         }
 512         obj->size = size;
 513
 514         switch (obj->type) {
 515         case OBJ_REF_DELTA:
 516                 hashcpy(ref_sha1, fill(20));
 517                 use(20);
 518                 break;
 519         case OBJ_OFS_DELTA:
 520                 p = fill(1);
 521                 c = *p;
 522                 use(1);
 523                 base_offset = c & 127;
 524                 while (c & 128) {
 525                         base_offset += 1;
 526                         if (!base_offset || MSB(base_offset, 7))
 527                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 528                         p = fill(1);
 529                         c = *p;
 530                         use(1);
 531                         base_offset = (base_offset << 7) + (c & 127);
 532                 }
 533                 *ofs_offset = obj->idx.offset - base_offset;
 534                 if (*ofs_offset <= 0 || *ofs_offset >= obj->idx.offset)
 535                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 536                 break;
 537         case OBJ_COMMIT:
 538         case OBJ_TREE:
 539         case OBJ_BLOB:
 540         case OBJ_TAG:
 541                 break;
 542         default:
 543                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 544         }
 545         obj->hdr_size = consumed_bytes - obj->idx.offset;
 546
 547         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 548         obj->idx.crc32 = input_crc32;
 549         return data;
 550 }
 551
 552 static void *unpack_data(struct object_entry *obj,
 553                          int (*consume)(const unsigned char *, unsigned long, void *),
 554                          void *cb_data)
 555 {
 556         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 557         off_t len = obj[1].idx.offset - from;
 558         unsigned char *data, *inbuf;
 559         git_zstream stream;
 560         int status;
 561
 562         data = xmallocz(consume ? 64*1024 : obj->size);
 563         inbuf = xmalloc((len < 64*1024) ? (int)len : 64*1024);
 564
 565         memset(&stream, 0, sizeof(stream));
 566         git_inflate_init(&stream);
 567         stream.next_out = data;
 568         stream.avail_out = consume ? 64*1024 : obj->size;
 569
 570         do {
 571                 ssize_t n = (len < 64*1024) ? (ssize_t)len : 64*1024;
 572                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 573                 if (n < 0)
 574                         die_errno(_("cannot pread pack file"));
 575                 if (!n)
 576                         die(Q_("premature end of pack file, %"PRIuMAX" byte missing",
 577                                "premature end of pack file, %"PRIuMAX" bytes missing",
 578                                (unsigned int)len),
 579                             (uintmax_t)len);
 580                 from += n;
 581                 len -= n;
 582                 stream.next_in = inbuf;
 583                 stream.avail_in = n;
 584                 if (!consume)
 585                         status = git_inflate(&stream, 0);
 586                 else {
 587                         do {
 588                                 status = git_inflate(&stream, 0);
 589                                 if (consume(data, stream.next_out - data, cb_data)) {
 590                                         free(inbuf);
 591                                         free(data);
 592                                         return NULL;
 593                                 }
 594                                 stream.next_out = data;
 595                                 stream.avail_out = 64*1024;
 596                         } while (status == Z_OK && stream.avail_in);
 597                 }
 598         } while (len && status == Z_OK && !stream.avail_in);
 599
 600         /* This has been inflated OK when first encountered, so... */
 601         if (status != Z_STREAM_END || stream.total_out != obj->size)
 602                 die(_("serious inflate inconsistency"));
 603
 604         git_inflate_end(&stream);
 605         free(inbuf);
 606         if (consume) {
 607                 free(data);
 608                 data = NULL;
 609         }
 610         return data;
 611 }
 612
 613 static void *get_data_from_pack(struct object_entry *obj)
 614 {
 615         return unpack_data(obj, NULL, NULL);
 616 }
 617
 618 static int compare_ofs_delta_bases(off_t offset1, off_t offset2,
 619                                    enum object_type type1,
 620                                    enum object_type type2)
 621 {
 622         int cmp = type1 - type2;
 623         if (cmp)
 624                 return cmp;
 625         return offset1 < offset2 ? -1 :
 626                offset1 > offset2 ?  1 :
 627                0;
 628 }
 629
 630 static int find_ofs_delta(const off_t offset, enum object_type type)
 631 {
 632         int first = 0, last = nr_ofs_deltas;
 633
 634         while (first < last) {
 635                 int next = (first + last) / 2;
 636                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 637                 int cmp;
 638
 639                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 640                                               type, objects[delta->obj_no].type);
 641                 if (!cmp)
 642                         return next;
 643                 if (cmp < 0) {
 644                         last = next;
 645                         continue;
 646                 }
 647                 first = next+1;
 648         }
 649         return -first-1;
 650 }
 651
 652 static void find_ofs_delta_children(off_t offset,
 653                                     int *first_index, int *last_index,
 654                                     enum object_type type)
 655 {
 656         int first = find_ofs_delta(offset, type);
 657         int last = first;
 658         int end = nr_ofs_deltas - 1;
 659
 660         if (first < 0) {
 661                 *first_index = 0;
 662                 *last_index = -1;
 663                 return;
 664         }
 665         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 666                 --first;
 667         while (last < end && ofs_deltas[last + 1].offset == offset)
 668                 ++last;
 669         *first_index = first;
 670         *last_index = last;
 671 }
 672
 673 static int compare_ref_delta_bases(const unsigned char *sha1,
 674                                    const unsigned char *sha2,
 675                                    enum object_type type1,
 676                                    enum object_type type2)
 677 {
 678         int cmp = type1 - type2;
 679         if (cmp)
 680                 return cmp;
 681         return hashcmp(sha1, sha2);
 682 }
 683
 684 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
 685 {
 686         int first = 0, last = nr_ref_deltas;
 687
 688         while (first < last) {
 689                 int next = (first + last) / 2;
 690                 struct ref_delta_entry *delta = &ref_deltas[next];
 691                 int cmp;
 692
 693                 cmp = compare_ref_delta_bases(sha1, delta->sha1,
 694                                               type, objects[delta->obj_no].type);
 695                 if (!cmp)
 696                         return next;
 697                 if (cmp < 0) {
 698                         last = next;
 699                         continue;
 700                 }
 701                 first = next+1;
 702         }
 703         return -first-1;
 704 }
 705
 706 static void find_ref_delta_children(const unsigned char *sha1,
 707                                     int *first_index, int *last_index,
 708                                     enum object_type type)
 709 {
 710         int first = find_ref_delta(sha1, type);
 711         int last = first;
 712         int end = nr_ref_deltas - 1;
 713
 714         if (first < 0) {
 715                 *first_index = 0;
 716                 *last_index = -1;
 717                 return;
 718         }
 719         while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
 720                 --first;
 721         while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
 722                 ++last;
 723         *first_index = first;
 724         *last_index = last;
 725 }
 726
 727 struct compare_data {
 728         struct object_entry *entry;
 729         struct git_istream *st;
 730         unsigned char *buf;
 731         unsigned long buf_size;
 732 };
 733
 734 static int compare_objects(const unsigned char *buf, unsigned long size,
 735                            void *cb_data)
 736 {
 737         struct compare_data *data = cb_data;
 738
 739         if (data->buf_size < size) {
 740                 free(data->buf);
 741                 data->buf = xmalloc(size);
 742                 data->buf_size = size;
 743         }
 744
 745         while (size) {
 746                 ssize_t len = read_istream(data->st, data->buf, size);
 747                 if (len == 0)
 748                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 749                             sha1_to_hex(data->entry->idx.sha1));
 750                 if (len < 0)
 751                         die(_("unable to read %s"),
 752                             sha1_to_hex(data->entry->idx.sha1));
 753                 if (memcmp(buf, data->buf, len))
 754                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 755                             sha1_to_hex(data->entry->idx.sha1));
 756                 size -= len;
 757                 buf += len;
 758         }
 759         return 0;
 760 }
 761
 762 static int check_collison(struct object_entry *entry)
 763 {
 764         struct compare_data data;
 765         enum object_type type;
 766         unsigned long size;
 767
 768         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 769                 return -1;
 770
 771         memset(&data, 0, sizeof(data));
 772         data.entry = entry;
 773         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 774         if (!data.st)
 775                 return -1;
 776         if (size != entry->size || type != entry->type)
 777                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 778                     sha1_to_hex(entry->idx.sha1));
 779         unpack_data(entry, compare_objects, &data);
 780         close_istream(data.st);
 781         free(data.buf);
 782         return 0;
 783 }
 784
 785 static void sha1_object(const void *data, struct object_entry *obj_entry,
 786                         unsigned long size, enum object_type type,
 787                         const unsigned char *sha1)
 788 {
 789         void *new_data = NULL;
 790         int collision_test_needed;
 791
 792         assert(data || obj_entry);
 793
 794         read_lock();
 795         collision_test_needed = has_sha1_file_with_flags(sha1, HAS_SHA1_QUICK);
 796         read_unlock();
 797
 798         if (collision_test_needed && !data) {
 799                 read_lock();
 800                 if (!check_collison(obj_entry))
 801                         collision_test_needed = 0;
 802                 read_unlock();
 803         }
 804         if (collision_test_needed) {
 805                 void *has_data;
 806                 enum object_type has_type;
 807                 unsigned long has_size;
 808                 read_lock();
 809                 has_type = sha1_object_info(sha1, &has_size);
 810                 if (has_type != type || has_size != size)
 811                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 812                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 813                 read_unlock();
 814                 if (!data)
 815                         data = new_data = get_data_from_pack(obj_entry);
 816                 if (!has_data)
 817                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 818                 if (size != has_size || type != has_type ||
 819                     memcmp(data, has_data, size) != 0)
 820                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 821                 free(has_data);
 822         }
 823
 824         if (strict) {
 825                 read_lock();
 826                 if (type == OBJ_BLOB) {
 827                         struct blob *blob = lookup_blob(sha1);
 828                         if (blob)
 829                                 blob->object.flags |= FLAG_CHECKED;
 830                         else
 831                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 832                 } else {
 833                         struct object *obj;
 834                         int eaten;
 835                         void *buf = (void *) data;
 836
 837                         assert(data && "data can only be NULL for large _blobs_");
 838
 839                         /*
 840                          * we do not need to free the memory here, as the
 841                          * buf is deleted by the caller.
 842                          */
 843                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 844                         if (!obj)
 845                                 die(_("invalid %s"), typename(type));
 846                         if (do_fsck_object &&
 847                             fsck_object(obj, buf, size, &fsck_options))
 848                                 die(_("Error in object"));
 849                         if (fsck_walk(obj, NULL, &fsck_options))
 850                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 851
 852                         if (obj->type == OBJ_TREE) {
 853                                 struct tree *item = (struct tree *) obj;
 854                                 item->buffer = NULL;
 855                                 obj->parsed = 0;
 856                         }
 857                         if (obj->type == OBJ_COMMIT) {
 858                                 struct commit *commit = (struct commit *) obj;
 859                                 if (detach_commit_buffer(commit, NULL) != data)
 860                                         die("BUG: parse_object_buffer transmogrified our buffer");
 861                         }
 862                         obj->flags |= FLAG_CHECKED;
 863                 }
 864                 read_unlock();
 865         }
 866
 867         free(new_data);
 868 }
 869
 870 /*
 871  * This function is part of find_unresolved_deltas(). There are two
 872  * walkers going in the opposite ways.
 873  *
 874  * The first one in find_unresolved_deltas() traverses down from
 875  * parent node to children, deflating nodes along the way. However,
 876  * memory for deflated nodes is limited by delta_base_cache_limit, so
 877  * at some point parent node's deflated content may be freed.
 878  *
 879  * The second walker is this function, which goes from current node up
 880  * to top parent if necessary to deflate the node. In normal
 881  * situation, its parent node would be already deflated, so it just
 882  * needs to apply delta.
 883  *
 884  * In the worst case scenario, parent node is no longer deflated because
 885  * we're running out of delta_base_cache_limit; we need to re-deflate
 886  * parents, possibly up to the top base.
 887  *
 888  * All deflated objects here are subject to be freed if we exceed
 889  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 890  * just need to make sure the last node is not freed.
 891  */
 892 static void *get_base_data(struct base_data *c)
 893 {
 894         if (!c->data) {
 895                 struct object_entry *obj = c->obj;
 896                 struct base_data **delta = NULL;
 897                 int delta_nr = 0, delta_alloc = 0;
 898
 899                 while (is_delta_type(c->obj->type) && !c->data) {
 900                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 901                         delta[delta_nr++] = c;
 902                         c = c->base;
 903                 }
 904                 if (!delta_nr) {
 905                         c->data = get_data_from_pack(obj);
 906                         c->size = obj->size;
 907                         get_thread_data()->base_cache_used += c->size;
 908                         prune_base_data(c);
 909                 }
 910                 for (; delta_nr > 0; delta_nr--) {
 911                         void *base, *raw;
 912                         c = delta[delta_nr - 1];
 913                         obj = c->obj;
 914                         base = get_base_data(c->base);
 915                         raw = get_data_from_pack(obj);
 916                         c->data = patch_delta(
 917                                 base, c->base->size,
 918                                 raw, obj->size,
 919                                 &c->size);
 920                         free(raw);
 921                         if (!c->data)
 922                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 923                         get_thread_data()->base_cache_used += c->size;
 924                         prune_base_data(c);
 925                 }
 926                 free(delta);
 927         }
 928         return c->data;
 929 }
 930
 931 static void resolve_delta(struct object_entry *delta_obj,
 932                           struct base_data *base, struct base_data *result)
 933 {
 934         void *base_data, *delta_data;
 935
 936         if (show_stat) {
 937                 int i = delta_obj - objects;
 938                 int j = base->obj - objects;
 939                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 940                 deepest_delta_lock();
 941                 if (deepest_delta < obj_stat[i].delta_depth)
 942                         deepest_delta = obj_stat[i].delta_depth;
 943                 deepest_delta_unlock();
 944                 obj_stat[i].base_object_no = j;
 945         }
 946         delta_data = get_data_from_pack(delta_obj);
 947         base_data = get_base_data(base);
 948         result->obj = delta_obj;
 949         result->data = patch_delta(base_data, base->size,
 950                                    delta_data, delta_obj->size, &result->size);
 951         free(delta_data);
 952         if (!result->data)
 953                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 954         hash_sha1_file(result->data, result->size,
 955                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 956         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 957                     delta_obj->idx.sha1);
 958         counter_lock();
 959         nr_resolved_deltas++;
 960         counter_unlock();
 961 }
 962
 963 /*
 964  * Standard boolean compare-and-swap: atomically check whether "*type" is
 965  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 966  * and return false.
 967  */
 968 static int compare_and_swap_type(signed char *type,
 969                                  enum object_type want,
 970                                  enum object_type set)
 971 {
 972         enum object_type old;
 973
 974         type_cas_lock();
 975         old = *type;
 976         if (old == want)
 977                 *type = set;
 978         type_cas_unlock();
 979
 980         return old == want;
 981 }
 982
 983 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 984                                                   struct base_data *prev_base)
 985 {
 986         if (base->ref_last == -1 && base->ofs_last == -1) {
 987                 find_ref_delta_children(base->obj->idx.sha1,
 988                                         &base->ref_first, &base->ref_last,
 989                                         OBJ_REF_DELTA);
 990
 991                 find_ofs_delta_children(base->obj->idx.offset,
 992                                         &base->ofs_first, &base->ofs_last,
 993                                         OBJ_OFS_DELTA);
 994
 995                 if (base->ref_last == -1 && base->ofs_last == -1) {
 996                         free(base->data);
 997                         return NULL;
 998                 }
 999
1000                 link_base_data(prev_base, base);
1001         }
1002
1003         if (base->ref_first <= base->ref_last) {
1004                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1005                 struct base_data *result = alloc_base_data();
1006
1007                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1008                                            base->obj->real_type))
1009                         die("BUG: child->real_type != OBJ_REF_DELTA");
1010
1011                 resolve_delta(child, base, result);
1012                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1013                         free_base_data(base);
1014
1015                 base->ref_first++;
1016                 return result;
1017         }
1018
1019         if (base->ofs_first <= base->ofs_last) {
1020                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1021                 struct base_data *result = alloc_base_data();
1022
1023                 assert(child->real_type == OBJ_OFS_DELTA);
1024                 child->real_type = base->obj->real_type;
1025                 resolve_delta(child, base, result);
1026                 if (base->ofs_first == base->ofs_last)
1027                         free_base_data(base);
1028
1029                 base->ofs_first++;
1030                 return result;
1031         }
1032
1033         unlink_base_data(base);
1034         return NULL;
1035 }
1036
1037 static void find_unresolved_deltas(struct base_data *base)
1038 {
1039         struct base_data *new_base, *prev_base = NULL;
1040         for (;;) {
1041                 new_base = find_unresolved_deltas_1(base, prev_base);
1042
1043                 if (new_base) {
1044                         prev_base = base;
1045                         base = new_base;
1046                 } else {
1047                         free(base);
1048                         base = prev_base;
1049                         if (!base)
1050                                 return;
1051                         prev_base = base->base;
1052                 }
1053         }
1054 }
1055
1056 static int compare_ofs_delta_entry(const void *a, const void *b)
1057 {
1058         const struct ofs_delta_entry *delta_a = a;
1059         const struct ofs_delta_entry *delta_b = b;
1060
1061         return delta_a->offset < delta_b->offset ? -1 :
1062                delta_a->offset > delta_b->offset ?  1 :
1063                0;
1064 }
1065
1066 static int compare_ref_delta_entry(const void *a, const void *b)
1067 {
1068         const struct ref_delta_entry *delta_a = a;
1069         const struct ref_delta_entry *delta_b = b;
1070
1071         return hashcmp(delta_a->sha1, delta_b->sha1);
1072 }
1073
1074 static void resolve_base(struct object_entry *obj)
1075 {
1076         struct base_data *base_obj = alloc_base_data();
1077         base_obj->obj = obj;
1078         base_obj->data = NULL;
1079         find_unresolved_deltas(base_obj);
1080 }
1081
1082 #ifndef NO_PTHREADS
1083 static void *threaded_second_pass(void *data)
1084 {
1085         set_thread_data(data);
1086         for (;;) {
1087                 int i;
1088                 counter_lock();
1089                 display_progress(progress, nr_resolved_deltas);
1090                 counter_unlock();
1091                 work_lock();
1092                 while (nr_dispatched < nr_objects &&
1093                        is_delta_type(objects[nr_dispatched].type))
1094                         nr_dispatched++;
1095                 if (nr_dispatched >= nr_objects) {
1096                         work_unlock();
1097                         break;
1098                 }
1099                 i = nr_dispatched++;
1100                 work_unlock();
1101
1102                 resolve_base(&objects[i]);
1103         }
1104         return NULL;
1105 }
1106 #endif
1107
1108 /*
1109  * First pass:
1110  * - find locations of all objects;
1111  * - calculate SHA1 of all non-delta objects;
1112  * - remember base (SHA1 or offset) for all deltas.
1113  */
1114 static void parse_pack_objects(unsigned char *sha1)
1115 {
1116         int i, nr_delays = 0;
1117         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1118         unsigned char ref_delta_sha1[20];
1119         struct stat st;
1120
1121         if (verbose)
1122                 progress = start_progress(
1123                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1124                                 nr_objects);
1125         for (i = 0; i < nr_objects; i++) {
1126                 struct object_entry *obj = &objects[i];
1127                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1128                                               ref_delta_sha1, obj->idx.sha1);
1129                 obj->real_type = obj->type;
1130                 if (obj->type == OBJ_OFS_DELTA) {
1131                         nr_ofs_deltas++;
1132                         ofs_delta->obj_no = i;
1133                         ofs_delta++;
1134                 } else if (obj->type == OBJ_REF_DELTA) {
1135                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1136                         hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1);
1137                         ref_deltas[nr_ref_deltas].obj_no = i;
1138                         nr_ref_deltas++;
1139                 } else if (!data) {
1140                         /* large blobs, check later */
1141                         obj->real_type = OBJ_BAD;
1142                         nr_delays++;
1143                 } else
1144                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1145                 free(data);
1146                 display_progress(progress, i+1);
1147         }
1148         objects[i].idx.offset = consumed_bytes;
1149         stop_progress(&progress);
1150
1151         /* Check pack integrity */
1152         flush();
1153         git_SHA1_Final(sha1, &input_ctx);
1154         if (hashcmp(fill(20), sha1))
1155                 die(_("pack is corrupted (SHA1 mismatch)"));
1156         use(20);
1157
1158         /* If input_fd is a file, we should have reached its end now. */
1159         if (fstat(input_fd, &st))
1160                 die_errno(_("cannot fstat packfile"));
1161         if (S_ISREG(st.st_mode) &&
1162                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1163                 die(_("pack has junk at the end"));
1164
1165         for (i = 0; i < nr_objects; i++) {
1166                 struct object_entry *obj = &objects[i];
1167                 if (obj->real_type != OBJ_BAD)
1168                         continue;
1169                 obj->real_type = obj->type;
1170                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1171                 nr_delays--;
1172         }
1173         if (nr_delays)
1174                 die(_("confusion beyond insanity in parse_pack_objects()"));
1175 }
1176
1177 /*
1178  * Second pass:
1179  * - for all non-delta objects, look if it is used as a base for
1180  *   deltas;
1181  * - if used as a base, uncompress the object and apply all deltas,
1182  *   recursively checking if the resulting object is used as a base
1183  *   for some more deltas.
1184  */
1185 static void resolve_deltas(void)
1186 {
1187         int i;
1188
1189         if (!nr_ofs_deltas && !nr_ref_deltas)
1190                 return;
1191
1192         /* Sort deltas by base SHA1/offset for fast searching */
1193         QSORT(ofs_deltas, nr_ofs_deltas, compare_ofs_delta_entry);
1194         QSORT(ref_deltas, nr_ref_deltas, compare_ref_delta_entry);
1195
1196         if (verbose || show_resolving_progress)
1197                 progress = start_progress(_("Resolving deltas"),
1198                                           nr_ref_deltas + nr_ofs_deltas);
1199
1200 #ifndef NO_PTHREADS
1201         nr_dispatched = 0;
1202         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1203                 init_thread();
1204                 for (i = 0; i < nr_threads; i++) {
1205                         int ret = pthread_create(&thread_data[i].thread, NULL,
1206                                                  threaded_second_pass, thread_data + i);
1207                         if (ret)
1208                                 die(_("unable to create thread: %s"),
1209                                     strerror(ret));
1210                 }
1211                 for (i = 0; i < nr_threads; i++)
1212                         pthread_join(thread_data[i].thread, NULL);
1213                 cleanup_thread();
1214                 return;
1215         }
1216 #endif
1217
1218         for (i = 0; i < nr_objects; i++) {
1219                 struct object_entry *obj = &objects[i];
1220
1221                 if (is_delta_type(obj->type))
1222                         continue;
1223                 resolve_base(obj);
1224                 display_progress(progress, nr_resolved_deltas);
1225         }
1226 }
1227
1228 /*
1229  * Third pass:
1230  * - append objects to convert thin pack to full pack if required
1231  * - write the final 20-byte SHA-1
1232  */
1233 static void fix_unresolved_deltas(struct sha1file *f);
1234 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1235 {
1236         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1237                 stop_progress(&progress);
1238                 /* Flush remaining pack final 20-byte SHA1. */
1239                 flush();
1240                 return;
1241         }
1242
1243         if (fix_thin_pack) {
1244                 struct sha1file *f;
1245                 unsigned char read_sha1[20], tail_sha1[20];
1246                 struct strbuf msg = STRBUF_INIT;
1247                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1248                 int nr_objects_initial = nr_objects;
1249                 if (nr_unresolved <= 0)
1250                         die(_("confusion beyond insanity"));
1251                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1252                 memset(objects + nr_objects + 1, 0,
1253                        nr_unresolved * sizeof(*objects));
1254                 f = sha1fd(output_fd, curr_pack);
1255                 fix_unresolved_deltas(f);
1256                 strbuf_addf(&msg, Q_("completed with %d local object",
1257                                      "completed with %d local objects",
1258                                      nr_objects - nr_objects_initial),
1259                             nr_objects - nr_objects_initial);
1260                 stop_progress_msg(&progress, msg.buf);
1261                 strbuf_release(&msg);
1262                 sha1close(f, tail_sha1, 0);
1263                 hashcpy(read_sha1, pack_sha1);
1264                 fixup_pack_header_footer(output_fd, pack_sha1,
1265                                          curr_pack, nr_objects,
1266                                          read_sha1, consumed_bytes-20);
1267                 if (hashcmp(read_sha1, tail_sha1) != 0)
1268                         die(_("Unexpected tail checksum for %s "
1269                               "(disk corruption?)"), curr_pack);
1270         }
1271         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1272                 die(Q_("pack has %d unresolved delta",
1273                        "pack has %d unresolved deltas",
1274                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1275                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1276 }
1277
1278 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1279 {
1280         git_zstream stream;
1281         int status;
1282         unsigned char outbuf[4096];
1283
1284         git_deflate_init(&stream, zlib_compression_level);
1285         stream.next_in = in;
1286         stream.avail_in = size;
1287
1288         do {
1289                 stream.next_out = outbuf;
1290                 stream.avail_out = sizeof(outbuf);
1291                 status = git_deflate(&stream, Z_FINISH);
1292                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1293         } while (status == Z_OK);
1294
1295         if (status != Z_STREAM_END)
1296                 die(_("unable to deflate appended object (%d)"), status);
1297         size = stream.total_out;
1298         git_deflate_end(&stream);
1299         return size;
1300 }
1301
1302 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1303                                const unsigned char *sha1, void *buf,
1304                                unsigned long size, enum object_type type)
1305 {
1306         struct object_entry *obj = &objects[nr_objects++];
1307         unsigned char header[10];
1308         unsigned long s = size;
1309         int n = 0;
1310         unsigned char c = (type << 4) | (s & 15);
1311         s >>= 4;
1312         while (s) {
1313                 header[n++] = c | 0x80;
1314                 c = s & 0x7f;
1315                 s >>= 7;
1316         }
1317         header[n++] = c;
1318         crc32_begin(f);
1319         sha1write(f, header, n);
1320         obj[0].size = size;
1321         obj[0].hdr_size = n;
1322         obj[0].type = type;
1323         obj[0].real_type = type;
1324         obj[1].idx.offset = obj[0].idx.offset + n;
1325         obj[1].idx.offset += write_compressed(f, buf, size);
1326         obj[0].idx.crc32 = crc32_end(f);
1327         sha1flush(f);
1328         hashcpy(obj->idx.sha1, sha1);
1329         return obj;
1330 }
1331
1332 static int delta_pos_compare(const void *_a, const void *_b)
1333 {
1334         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1335         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1336         return a->obj_no - b->obj_no;
1337 }
1338
1339 static void fix_unresolved_deltas(struct sha1file *f)
1340 {
1341         struct ref_delta_entry **sorted_by_pos;
1342         int i;
1343
1344         /*
1345          * Since many unresolved deltas may well be themselves base objects
1346          * for more unresolved deltas, we really want to include the
1347          * smallest number of base objects that would cover as much delta
1348          * as possible by picking the
1349          * trunc deltas first, allowing for other deltas to resolve without
1350          * additional base objects.  Since most base objects are to be found
1351          * before deltas depending on them, a good heuristic is to start
1352          * resolving deltas in the same order as their position in the pack.
1353          */
1354         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1355         for (i = 0; i < nr_ref_deltas; i++)
1356                 sorted_by_pos[i] = &ref_deltas[i];
1357         QSORT(sorted_by_pos, nr_ref_deltas, delta_pos_compare);
1358
1359         for (i = 0; i < nr_ref_deltas; i++) {
1360                 struct ref_delta_entry *d = sorted_by_pos[i];
1361                 enum object_type type;
1362                 struct base_data *base_obj = alloc_base_data();
1363
1364                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1365                         continue;
1366                 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1367                 if (!base_obj->data)
1368                         continue;
1369
1370                 if (check_sha1_signature(d->sha1, base_obj->data,
1371                                 base_obj->size, typename(type)))
1372                         die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1373                 base_obj->obj = append_obj_to_pack(f, d->sha1,
1374                                         base_obj->data, base_obj->size, type);
1375                 find_unresolved_deltas(base_obj);
1376                 display_progress(progress, nr_resolved_deltas);
1377         }
1378         free(sorted_by_pos);
1379 }
1380
1381 static void final(const char *final_pack_name, const char *curr_pack_name,
1382                   const char *final_index_name, const char *curr_index_name,
1383                   const char *keep_name, const char *keep_msg,
1384                   unsigned char *sha1)
1385 {
1386         const char *report = "pack";
1387         char name[PATH_MAX];
1388         int err;
1389
1390         if (!from_stdin) {
1391                 close(input_fd);
1392         } else {
1393                 fsync_or_die(output_fd, curr_pack_name);
1394                 err = close(output_fd);
1395                 if (err)
1396                         die_errno(_("error while closing pack file"));
1397         }
1398
1399         if (keep_msg) {
1400                 int keep_fd, keep_msg_len = strlen(keep_msg);
1401
1402                 if (!keep_name)
1403                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1404                 else
1405                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1406
1407                 if (keep_fd < 0) {
1408                         if (errno != EEXIST)
1409                                 die_errno(_("cannot write keep file '%s'"),
1410                                           keep_name ? keep_name : name);
1411                 } else {
1412                         if (keep_msg_len > 0) {
1413                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1414                                 write_or_die(keep_fd, "\n", 1);
1415                         }
1416                         if (close(keep_fd) != 0)
1417                                 die_errno(_("cannot close written keep file '%s'"),
1418                                           keep_name ? keep_name : name);
1419                         report = "keep";
1420                 }
1421         }
1422
1423         if (final_pack_name != curr_pack_name) {
1424                 if (!final_pack_name) {
1425                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1426                                  get_object_directory(), sha1_to_hex(sha1));
1427                         final_pack_name = name;
1428                 }
1429                 if (finalize_object_file(curr_pack_name, final_pack_name))
1430                         die(_("cannot store pack file"));
1431         } else if (from_stdin)
1432                 chmod(final_pack_name, 0444);
1433
1434         if (final_index_name != curr_index_name) {
1435                 if (!final_index_name) {
1436                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1437                                  get_object_directory(), sha1_to_hex(sha1));
1438                         final_index_name = name;
1439                 }
1440                 if (finalize_object_file(curr_index_name, final_index_name))
1441                         die(_("cannot store index file"));
1442         } else
1443                 chmod(final_index_name, 0444);
1444
1445         if (!from_stdin) {
1446                 printf("%s\n", sha1_to_hex(sha1));
1447         } else {
1448                 char buf[48];
1449                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1450                                    report, sha1_to_hex(sha1));
1451                 write_or_die(1, buf, len);
1452
1453                 /*
1454                  * Let's just mimic git-unpack-objects here and write
1455                  * the last part of the input buffer to stdout.
1456                  */
1457                 while (input_len) {
1458                         err = xwrite(1, input_buffer + input_offset, input_len);
1459                         if (err <= 0)
1460                                 break;
1461                         input_len -= err;
1462                         input_offset += err;
1463                 }
1464         }
1465 }
1466
1467 static int git_index_pack_config(const char *k, const char *v, void *cb)
1468 {
1469         struct pack_idx_option *opts = cb;
1470
1471         if (!strcmp(k, "pack.indexversion")) {
1472                 opts->version = git_config_int(k, v);
1473                 if (opts->version > 2)
1474                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1475                 return 0;
1476         }
1477         if (!strcmp(k, "pack.threads")) {
1478                 nr_threads = git_config_int(k, v);
1479                 if (nr_threads < 0)
1480                         die(_("invalid number of threads specified (%d)"),
1481                             nr_threads);
1482 #ifdef NO_PTHREADS
1483                 if (nr_threads != 1)
1484                         warning(_("no threads support, ignoring %s"), k);
1485                 nr_threads = 1;
1486 #endif
1487                 return 0;
1488         }
1489         return git_default_config(k, v, cb);
1490 }
1491
1492 static int cmp_uint32(const void *a_, const void *b_)
1493 {
1494         uint32_t a = *((uint32_t *)a_);
1495         uint32_t b = *((uint32_t *)b_);
1496
1497         return (a < b) ? -1 : (a != b);
1498 }
1499
1500 static void read_v2_anomalous_offsets(struct packed_git *p,
1501                                       struct pack_idx_option *opts)
1502 {
1503         const uint32_t *idx1, *idx2;
1504         uint32_t i;
1505
1506         /* The address of the 4-byte offset table */
1507         idx1 = (((const uint32_t *)p->index_data)
1508                 + 2 /* 8-byte header */
1509                 + 256 /* fan out */
1510                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1511                 + p->num_objects /* CRC32 table */
1512                 );
1513
1514         /* The address of the 8-byte offset table */
1515         idx2 = idx1 + p->num_objects;
1516
1517         for (i = 0; i < p->num_objects; i++) {
1518                 uint32_t off = ntohl(idx1[i]);
1519                 if (!(off & 0x80000000))
1520                         continue;
1521                 off = off & 0x7fffffff;
1522                 check_pack_index_ptr(p, &idx2[off * 2]);
1523                 if (idx2[off * 2])
1524                         continue;
1525                 /*
1526                  * The real offset is ntohl(idx2[off * 2]) in high 4
1527                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1528                  * octets.  But idx2[off * 2] is Zero!!!
1529                  */
1530                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1531                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1532         }
1533
1534         QSORT(opts->anomaly, opts->anomaly_nr, cmp_uint32);
1535 }
1536
1537 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1538 {
1539         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1540
1541         if (!p)
1542                 die(_("Cannot open existing pack file '%s'"), pack_name);
1543         if (open_pack_index(p))
1544                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1545
1546         /* Read the attributes from the existing idx file */
1547         opts->version = p->index_version;
1548
1549         if (opts->version == 2)
1550                 read_v2_anomalous_offsets(p, opts);
1551
1552         /*
1553          * Get rid of the idx file as we do not need it anymore.
1554          * NEEDSWORK: extract this bit from free_pack_by_name() in
1555          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1556          * know we haven't installed this pack (hence we never have
1557          * read anything from it).
1558          */
1559         close_pack_index(p);
1560         free(p);
1561 }
1562
1563 static void show_pack_info(int stat_only)
1564 {
1565         int i, baseobjects = nr_objects - nr_ref_deltas - nr_ofs_deltas;
1566         unsigned long *chain_histogram = NULL;
1567
1568         if (deepest_delta)
1569                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1570
1571         for (i = 0; i < nr_objects; i++) {
1572                 struct object_entry *obj = &objects[i];
1573
1574                 if (is_delta_type(obj->type))
1575                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1576                 if (stat_only)
1577                         continue;
1578                 printf("%s %-6s %lu %lu %"PRIuMAX,
1579                        sha1_to_hex(obj->idx.sha1),
1580                        typename(obj->real_type), obj->size,
1581                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1582                        (uintmax_t)obj->idx.offset);
1583                 if (is_delta_type(obj->type)) {
1584                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1585                         printf(" %u %s", obj_stat[i].delta_depth, sha1_to_hex(bobj->idx.sha1));
1586                 }
1587                 putchar('\n');
1588         }
1589
1590         if (baseobjects)
1591                 printf_ln(Q_("non delta: %d object",
1592                              "non delta: %d objects",
1593                              baseobjects),
1594                           baseobjects);
1595         for (i = 0; i < deepest_delta; i++) {
1596                 if (!chain_histogram[i])
1597                         continue;
1598                 printf_ln(Q_("chain length = %d: %lu object",
1599                              "chain length = %d: %lu objects",
1600                              chain_histogram[i]),
1601                           i + 1,
1602                           chain_histogram[i]);
1603         }
1604 }
1605
1606 static const char *derive_filename(const char *pack_name, const char *suffix,
1607                                    struct strbuf *buf)
1608 {
1609         size_t len;
1610         if (!strip_suffix(pack_name, ".pack", &len))
1611                 die(_("packfile name '%s' does not end with '.pack'"),
1612                     pack_name);
1613         strbuf_add(buf, pack_name, len);
1614         strbuf_addstr(buf, suffix);
1615         return buf->buf;
1616 }
1617
1618 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1619 {
1620         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1621         const char *curr_index;
1622         const char *index_name = NULL, *pack_name = NULL;
1623         const char *keep_name = NULL, *keep_msg = NULL;
1624         struct strbuf index_name_buf = STRBUF_INIT,
1625                       keep_name_buf = STRBUF_INIT;
1626         struct pack_idx_entry **idx_objects;
1627         struct pack_idx_option opts;
1628         unsigned char pack_sha1[20];
1629         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1630         int report_end_of_input = 0;
1631
1632         if (argc == 2 && !strcmp(argv[1], "-h"))
1633                 usage(index_pack_usage);
1634
1635         check_replace_refs = 0;
1636         fsck_options.walk = mark_link;
1637
1638         reset_pack_idx_option(&opts);
1639         git_config(git_index_pack_config, &opts);
1640         if (prefix && chdir(prefix))
1641                 die(_("Cannot come back to cwd"));
1642
1643         for (i = 1; i < argc; i++) {
1644                 const char *arg = argv[i];
1645
1646                 if (*arg == '-') {
1647                         if (!strcmp(arg, "--stdin")) {
1648                                 from_stdin = 1;
1649                         } else if (!strcmp(arg, "--fix-thin")) {
1650                                 fix_thin_pack = 1;
1651                         } else if (!strcmp(arg, "--strict")) {
1652                                 strict = 1;
1653                                 do_fsck_object = 1;
1654                         } else if (skip_prefix(arg, "--strict=", &arg)) {
1655                                 strict = 1;
1656                                 do_fsck_object = 1;
1657                                 fsck_set_msg_types(&fsck_options, arg);
1658                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1659                                 strict = 1;
1660                                 check_self_contained_and_connected = 1;
1661                         } else if (!strcmp(arg, "--verify")) {
1662                                 verify = 1;
1663                         } else if (!strcmp(arg, "--verify-stat")) {
1664                                 verify = 1;
1665                                 show_stat = 1;
1666                         } else if (!strcmp(arg, "--verify-stat-only")) {
1667                                 verify = 1;
1668                                 show_stat = 1;
1669                                 stat_only = 1;
1670                         } else if (!strcmp(arg, "--keep")) {
1671                                 keep_msg = "";
1672                         } else if (starts_with(arg, "--keep=")) {
1673                                 keep_msg = arg + 7;
1674                         } else if (starts_with(arg, "--threads=")) {
1675                                 char *end;
1676                                 nr_threads = strtoul(arg+10, &end, 0);
1677                                 if (!arg[10] || *end || nr_threads < 0)
1678                                         usage(index_pack_usage);
1679 #ifdef NO_PTHREADS
1680                                 if (nr_threads != 1)
1681                                         warning(_("no threads support, "
1682                                                   "ignoring %s"), arg);
1683                                 nr_threads = 1;
1684 #endif
1685                         } else if (starts_with(arg, "--pack_header=")) {
1686                                 struct pack_header *hdr;
1687                                 char *c;
1688
1689                                 hdr = (struct pack_header *)input_buffer;
1690                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1691                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1692                                 if (*c != ',')
1693                                         die(_("bad %s"), arg);
1694                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1695                                 if (*c)
1696                                         die(_("bad %s"), arg);
1697                                 input_len = sizeof(*hdr);
1698                         } else if (!strcmp(arg, "-v")) {
1699                                 verbose = 1;
1700                         } else if (!strcmp(arg, "--show-resolving-progress")) {
1701                                 show_resolving_progress = 1;
1702                         } else if (!strcmp(arg, "--report-end-of-input")) {
1703                                 report_end_of_input = 1;
1704                         } else if (!strcmp(arg, "-o")) {
1705                                 if (index_name || (i+1) >= argc)
1706                                         usage(index_pack_usage);
1707                                 index_name = argv[++i];
1708                         } else if (starts_with(arg, "--index-version=")) {
1709                                 char *c;
1710                                 opts.version = strtoul(arg + 16, &c, 10);
1711                                 if (opts.version > 2)
1712                                         die(_("bad %s"), arg);
1713                                 if (*c == ',')
1714                                         opts.off32_limit = strtoul(c+1, &c, 0);
1715                                 if (*c || opts.off32_limit & 0x80000000)
1716                                         die(_("bad %s"), arg);
1717                         } else if (skip_prefix(arg, "--max-input-size=", &arg)) {
1718                                 max_input_size = strtoumax(arg, NULL, 10);
1719                         } else
1720                                 usage(index_pack_usage);
1721                         continue;
1722                 }
1723
1724                 if (pack_name)
1725                         usage(index_pack_usage);
1726                 pack_name = arg;
1727         }
1728
1729         if (!pack_name && !from_stdin)
1730                 usage(index_pack_usage);
1731         if (fix_thin_pack && !from_stdin)
1732                 die(_("--fix-thin cannot be used without --stdin"));
1733         if (from_stdin && !startup_info->have_repository)
1734                 die(_("--stdin requires a git repository"));
1735         if (!index_name && pack_name)
1736                 index_name = derive_filename(pack_name, ".idx", &index_name_buf);
1737         if (keep_msg && !keep_name && pack_name)
1738                 keep_name = derive_filename(pack_name, ".keep", &keep_name_buf);
1739
1740         if (verify) {
1741                 if (!index_name)
1742                         die(_("--verify with no packfile name given"));
1743                 read_idx_option(&opts, index_name);
1744                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1745         }
1746         if (strict)
1747                 opts.flags |= WRITE_IDX_STRICT;
1748
1749 #ifndef NO_PTHREADS
1750         if (!nr_threads) {
1751                 nr_threads = online_cpus();
1752                 /* An experiment showed that more threads does not mean faster */
1753                 if (nr_threads > 3)
1754                         nr_threads = 3;
1755         }
1756 #endif
1757
1758         curr_pack = open_pack_file(pack_name);
1759         parse_pack_header();
1760         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1761         if (show_stat)
1762                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1763         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1764         parse_pack_objects(pack_sha1);
1765         if (report_end_of_input)
1766                 write_in_full(2, "\0", 1);
1767         resolve_deltas();
1768         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1769         free(ofs_deltas);
1770         free(ref_deltas);
1771         if (strict)
1772                 foreign_nr = check_objects();
1773
1774         if (show_stat)
1775                 show_pack_info(stat_only);
1776
1777         ALLOC_ARRAY(idx_objects, nr_objects);
1778         for (i = 0; i < nr_objects; i++)
1779                 idx_objects[i] = &objects[i].idx;
1780         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1781         free(idx_objects);
1782
1783         if (!verify)
1784                 final(pack_name, curr_pack,
1785                       index_name, curr_index,
1786                       keep_name, keep_msg,
1787                       pack_sha1);
1788         else
1789                 close(input_fd);
1790         free(objects);
1791         strbuf_release(&index_name_buf);
1792         strbuf_release(&keep_name_buf);
1793         if (pack_name == NULL)
1794                 free((void *) curr_pack);
1795         if (index_name == NULL)
1796                 free((void *) curr_index);
1797
1798         /*
1799          * Let the caller know this pack is not self contained
1800          */
1801         if (check_self_contained_and_connected && foreign_nr)
1802                 return 1;
1803
1804         return 0;
1805 }