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 union delta_base {
  32         unsigned char sha1[20];
  33         off_t offset;
  34 };
  35
  36 struct base_data {
  37         struct base_data *base;
  38         struct base_data *child;
  39         struct object_entry *obj;
  40         void *data;
  41         unsigned long size;
  42         int ref_first, ref_last;
  43         int ofs_first, ofs_last;
  44 };
  45
  46 struct thread_local {
  47 #ifndef NO_PTHREADS
  48         pthread_t thread;
  49 #endif
  50         struct base_data *base_cache;
  51         size_t base_cache_used;
  52         int pack_fd;
  53 };
  54
  55 /*
  56  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  57  * to memcmp() only the first 20 bytes.
  58  */
  59 #define UNION_BASE_SZ   20
  60
  61 #define FLAG_LINK (1u<<20)
  62 #define FLAG_CHECKED (1u<<21)
  63
  64 struct delta_entry {
  65         union delta_base base;
  66         int obj_no;
  67 };
  68
  69 static struct object_entry *objects;
  70 static struct object_stat *obj_stat;
  71 static struct delta_entry *deltas;
  72 static struct thread_local nothread_data;
  73 static int nr_objects;
  74 static int nr_deltas;
  75 static int nr_resolved_deltas;
  76 static int nr_threads;
  77
  78 static int from_stdin;
  79 static int strict;
  80 static int do_fsck_object;
  81 static int verbose;
  82 static int show_stat;
  83 static int check_self_contained_and_connected;
  84
  85 static struct progress *progress;
  86
  87 /* We always read in 4kB chunks. */
  88 static unsigned char input_buffer[4096];
  89 static unsigned int input_offset, input_len;
  90 static off_t consumed_bytes;
  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)
 199 {
 200         if (!obj)
 201                 return -1;
 202
 203         if (type != OBJ_ANY && obj->type != type)
 204                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 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->sha1, &size);
 223                 if (type <= 0)
 224                         die(_("did not receive expected object %s"),
 225                               sha1_to_hex(obj->sha1));
 226                 if (type != obj->type)
 227                         die(_("object %s: expected type %s, found %s"),
 228                             sha1_to_hex(obj->sha1),
 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 }
 302
 303 static const char *open_pack_file(const char *pack_name)
 304 {
 305         if (from_stdin) {
 306                 input_fd = 0;
 307                 if (!pack_name) {
 308                         static char tmp_file[PATH_MAX];
 309                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 310                                                 "pack/tmp_pack_XXXXXX");
 311                         pack_name = xstrdup(tmp_file);
 312                 } else
 313                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 314                 if (output_fd < 0)
 315                         die_errno(_("unable to create '%s'"), pack_name);
 316                 nothread_data.pack_fd = output_fd;
 317         } else {
 318                 input_fd = open(pack_name, O_RDONLY);
 319                 if (input_fd < 0)
 320                         die_errno(_("cannot open packfile '%s'"), pack_name);
 321                 output_fd = -1;
 322                 nothread_data.pack_fd = input_fd;
 323         }
 324         git_SHA1_Init(&input_ctx);
 325         return pack_name;
 326 }
 327
 328 static void parse_pack_header(void)
 329 {
 330         struct pack_header *hdr = fill(sizeof(struct pack_header));
 331
 332         /* Header consistency check */
 333         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 334                 die(_("pack signature mismatch"));
 335         if (!pack_version_ok(hdr->hdr_version))
 336                 die(_("pack version %"PRIu32" unsupported"),
 337                         ntohl(hdr->hdr_version));
 338
 339         nr_objects = ntohl(hdr->hdr_entries);
 340         use(sizeof(struct pack_header));
 341 }
 342
 343 static NORETURN void bad_object(unsigned long offset, const char *format,
 344                        ...) __attribute__((format (printf, 2, 3)));
 345
 346 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 347 {
 348         va_list params;
 349         char buf[1024];
 350
 351         va_start(params, format);
 352         vsnprintf(buf, sizeof(buf), format, params);
 353         va_end(params);
 354         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 355 }
 356
 357 static inline struct thread_local *get_thread_data(void)
 358 {
 359 #ifndef NO_PTHREADS
 360         if (threads_active)
 361                 return pthread_getspecific(key);
 362         assert(!threads_active &&
 363                "This should only be reached when all threads are gone");
 364 #endif
 365         return &nothread_data;
 366 }
 367
 368 #ifndef NO_PTHREADS
 369 static void set_thread_data(struct thread_local *data)
 370 {
 371         if (threads_active)
 372                 pthread_setspecific(key, data);
 373 }
 374 #endif
 375
 376 static struct base_data *alloc_base_data(void)
 377 {
 378         struct base_data *base = xcalloc(1, sizeof(struct base_data));
 379         base->ref_last = -1;
 380         base->ofs_last = -1;
 381         return base;
 382 }
 383
 384 static void free_base_data(struct base_data *c)
 385 {
 386         if (c->data) {
 387                 free(c->data);
 388                 c->data = NULL;
 389                 get_thread_data()->base_cache_used -= c->size;
 390         }
 391 }
 392
 393 static void prune_base_data(struct base_data *retain)
 394 {
 395         struct base_data *b;
 396         struct thread_local *data = get_thread_data();
 397         for (b = data->base_cache;
 398              data->base_cache_used > delta_base_cache_limit && b;
 399              b = b->child) {
 400                 if (b->data && b != retain)
 401                         free_base_data(b);
 402         }
 403 }
 404
 405 static void link_base_data(struct base_data *base, struct base_data *c)
 406 {
 407         if (base)
 408                 base->child = c;
 409         else
 410                 get_thread_data()->base_cache = c;
 411
 412         c->base = base;
 413         c->child = NULL;
 414         if (c->data)
 415                 get_thread_data()->base_cache_used += c->size;
 416         prune_base_data(c);
 417 }
 418
 419 static void unlink_base_data(struct base_data *c)
 420 {
 421         struct base_data *base = c->base;
 422         if (base)
 423                 base->child = NULL;
 424         else
 425                 get_thread_data()->base_cache = NULL;
 426         free_base_data(c);
 427 }
 428
 429 static int is_delta_type(enum object_type type)
 430 {
 431         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 432 }
 433
 434 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 435                                enum object_type type, unsigned char *sha1)
 436 {
 437         static char fixed_buf[8192];
 438         int status;
 439         git_zstream stream;
 440         void *buf;
 441         git_SHA_CTX c;
 442         char hdr[32];
 443         int hdrlen;
 444
 445         if (!is_delta_type(type)) {
 446                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 447                 git_SHA1_Init(&c);
 448                 git_SHA1_Update(&c, hdr, hdrlen);
 449         } else
 450                 sha1 = NULL;
 451         if (type == OBJ_BLOB && size > big_file_threshold)
 452                 buf = fixed_buf;
 453         else
 454                 buf = xmallocz(size);
 455
 456         memset(&stream, 0, sizeof(stream));
 457         git_inflate_init(&stream);
 458         stream.next_out = buf;
 459         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 460
 461         do {
 462                 unsigned char *last_out = stream.next_out;
 463                 stream.next_in = fill(1);
 464                 stream.avail_in = input_len;
 465                 status = git_inflate(&stream, 0);
 466                 use(input_len - stream.avail_in);
 467                 if (sha1)
 468                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 469                 if (buf == fixed_buf) {
 470                         stream.next_out = buf;
 471                         stream.avail_out = sizeof(fixed_buf);
 472                 }
 473         } while (status == Z_OK);
 474         if (stream.total_out != size || status != Z_STREAM_END)
 475                 bad_object(offset, _("inflate returned %d"), status);
 476         git_inflate_end(&stream);
 477         if (sha1)
 478                 git_SHA1_Final(sha1, &c);
 479         return buf == fixed_buf ? NULL : buf;
 480 }
 481
 482 static void *unpack_raw_entry(struct object_entry *obj,
 483                               union delta_base *delta_base,
 484                               unsigned char *sha1)
 485 {
 486         unsigned char *p;
 487         unsigned long size, c;
 488         off_t base_offset;
 489         unsigned shift;
 490         void *data;
 491
 492         obj->idx.offset = consumed_bytes;
 493         input_crc32 = crc32(0, NULL, 0);
 494
 495         p = fill(1);
 496         c = *p;
 497         use(1);
 498         obj->type = (c >> 4) & 7;
 499         size = (c & 15);
 500         shift = 4;
 501         while (c & 0x80) {
 502                 p = fill(1);
 503                 c = *p;
 504                 use(1);
 505                 size += (c & 0x7f) << shift;
 506                 shift += 7;
 507         }
 508         obj->size = size;
 509
 510         switch (obj->type) {
 511         case OBJ_REF_DELTA:
 512                 hashcpy(delta_base->sha1, fill(20));
 513                 use(20);
 514                 break;
 515         case OBJ_OFS_DELTA:
 516                 memset(delta_base, 0, sizeof(*delta_base));
 517                 p = fill(1);
 518                 c = *p;
 519                 use(1);
 520                 base_offset = c & 127;
 521                 while (c & 128) {
 522                         base_offset += 1;
 523                         if (!base_offset || MSB(base_offset, 7))
 524                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 525                         p = fill(1);
 526                         c = *p;
 527                         use(1);
 528                         base_offset = (base_offset << 7) + (c & 127);
 529                 }
 530                 delta_base->offset = obj->idx.offset - base_offset;
 531                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 532                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 533                 break;
 534         case OBJ_COMMIT:
 535         case OBJ_TREE:
 536         case OBJ_BLOB:
 537         case OBJ_TAG:
 538                 break;
 539         default:
 540                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 541         }
 542         obj->hdr_size = consumed_bytes - obj->idx.offset;
 543
 544         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 545         obj->idx.crc32 = input_crc32;
 546         return data;
 547 }
 548
 549 static void *unpack_data(struct object_entry *obj,
 550                          int (*consume)(const unsigned char *, unsigned long, void *),
 551                          void *cb_data)
 552 {
 553         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 554         unsigned long len = obj[1].idx.offset - from;
 555         unsigned char *data, *inbuf;
 556         git_zstream stream;
 557         int status;
 558
 559         data = xmallocz(consume ? 64*1024 : obj->size);
 560         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 561
 562         memset(&stream, 0, sizeof(stream));
 563         git_inflate_init(&stream);
 564         stream.next_out = data;
 565         stream.avail_out = consume ? 64*1024 : obj->size;
 566
 567         do {
 568                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 569                 n = xpread(get_thread_data()->pack_fd, inbuf, n, from);
 570                 if (n < 0)
 571                         die_errno(_("cannot pread pack file"));
 572                 if (!n)
 573                         die(Q_("premature end of pack file, %lu byte missing",
 574                                "premature end of pack file, %lu bytes missing",
 575                                len),
 576                             len);
 577                 from += n;
 578                 len -= n;
 579                 stream.next_in = inbuf;
 580                 stream.avail_in = n;
 581                 if (!consume)
 582                         status = git_inflate(&stream, 0);
 583                 else {
 584                         do {
 585                                 status = git_inflate(&stream, 0);
 586                                 if (consume(data, stream.next_out - data, cb_data)) {
 587                                         free(inbuf);
 588                                         free(data);
 589                                         return NULL;
 590                                 }
 591                                 stream.next_out = data;
 592                                 stream.avail_out = 64*1024;
 593                         } while (status == Z_OK && stream.avail_in);
 594                 }
 595         } while (len && status == Z_OK && !stream.avail_in);
 596
 597         /* This has been inflated OK when first encountered, so... */
 598         if (status != Z_STREAM_END || stream.total_out != obj->size)
 599                 die(_("serious inflate inconsistency"));
 600
 601         git_inflate_end(&stream);
 602         free(inbuf);
 603         if (consume) {
 604                 free(data);
 605                 data = NULL;
 606         }
 607         return data;
 608 }
 609
 610 static void *get_data_from_pack(struct object_entry *obj)
 611 {
 612         return unpack_data(obj, NULL, NULL);
 613 }
 614
 615 static int compare_delta_bases(const union delta_base *base1,
 616                                const union delta_base *base2,
 617                                enum object_type type1,
 618                                enum object_type type2)
 619 {
 620         int cmp = type1 - type2;
 621         if (cmp)
 622                 return cmp;
 623         return memcmp(base1, base2, UNION_BASE_SZ);
 624 }
 625
 626 static int find_delta(const union delta_base *base, enum object_type type)
 627 {
 628         int first = 0, last = nr_deltas;
 629
 630         while (first < last) {
 631                 int next = (first + last) / 2;
 632                 struct delta_entry *delta = &deltas[next];
 633                 int cmp;
 634
 635                 cmp = compare_delta_bases(base, &delta->base,
 636                                           type, objects[delta->obj_no].type);
 637                 if (!cmp)
 638                         return next;
 639                 if (cmp < 0) {
 640                         last = next;
 641                         continue;
 642                 }
 643                 first = next+1;
 644         }
 645         return -first-1;
 646 }
 647
 648 static void find_delta_children(const union delta_base *base,
 649                                 int *first_index, int *last_index,
 650                                 enum object_type type)
 651 {
 652         int first = find_delta(base, type);
 653         int last = first;
 654         int end = nr_deltas - 1;
 655
 656         if (first < 0) {
 657                 *first_index = 0;
 658                 *last_index = -1;
 659                 return;
 660         }
 661         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 662                 --first;
 663         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 664                 ++last;
 665         *first_index = first;
 666         *last_index = last;
 667 }
 668
 669 struct compare_data {
 670         struct object_entry *entry;
 671         struct git_istream *st;
 672         unsigned char *buf;
 673         unsigned long buf_size;
 674 };
 675
 676 static int compare_objects(const unsigned char *buf, unsigned long size,
 677                            void *cb_data)
 678 {
 679         struct compare_data *data = cb_data;
 680
 681         if (data->buf_size < size) {
 682                 free(data->buf);
 683                 data->buf = xmalloc(size);
 684                 data->buf_size = size;
 685         }
 686
 687         while (size) {
 688                 ssize_t len = read_istream(data->st, data->buf, size);
 689                 if (len == 0)
 690                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 691                             sha1_to_hex(data->entry->idx.sha1));
 692                 if (len < 0)
 693                         die(_("unable to read %s"),
 694                             sha1_to_hex(data->entry->idx.sha1));
 695                 if (memcmp(buf, data->buf, len))
 696                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 697                             sha1_to_hex(data->entry->idx.sha1));
 698                 size -= len;
 699                 buf += len;
 700         }
 701         return 0;
 702 }
 703
 704 static int check_collison(struct object_entry *entry)
 705 {
 706         struct compare_data data;
 707         enum object_type type;
 708         unsigned long size;
 709
 710         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 711                 return -1;
 712
 713         memset(&data, 0, sizeof(data));
 714         data.entry = entry;
 715         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 716         if (!data.st)
 717                 return -1;
 718         if (size != entry->size || type != entry->type)
 719                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 720                     sha1_to_hex(entry->idx.sha1));
 721         unpack_data(entry, compare_objects, &data);
 722         close_istream(data.st);
 723         free(data.buf);
 724         return 0;
 725 }
 726
 727 static void sha1_object(const void *data, struct object_entry *obj_entry,
 728                         unsigned long size, enum object_type type,
 729                         const unsigned char *sha1)
 730 {
 731         void *new_data = NULL;
 732         int collision_test_needed;
 733
 734         assert(data || obj_entry);
 735
 736         read_lock();
 737         collision_test_needed = has_sha1_file(sha1);
 738         read_unlock();
 739
 740         if (collision_test_needed && !data) {
 741                 read_lock();
 742                 if (!check_collison(obj_entry))
 743                         collision_test_needed = 0;
 744                 read_unlock();
 745         }
 746         if (collision_test_needed) {
 747                 void *has_data;
 748                 enum object_type has_type;
 749                 unsigned long has_size;
 750                 read_lock();
 751                 has_type = sha1_object_info(sha1, &has_size);
 752                 if (has_type != type || has_size != size)
 753                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 754                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 755                 read_unlock();
 756                 if (!data)
 757                         data = new_data = get_data_from_pack(obj_entry);
 758                 if (!has_data)
 759                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 760                 if (size != has_size || type != has_type ||
 761                     memcmp(data, has_data, size) != 0)
 762                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 763                 free(has_data);
 764         }
 765
 766         if (strict) {
 767                 read_lock();
 768                 if (type == OBJ_BLOB) {
 769                         struct blob *blob = lookup_blob(sha1);
 770                         if (blob)
 771                                 blob->object.flags |= FLAG_CHECKED;
 772                         else
 773                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 774                 } else {
 775                         struct object *obj;
 776                         int eaten;
 777                         void *buf = (void *) data;
 778
 779                         assert(data && "data can only be NULL for large _blobs_");
 780
 781                         /*
 782                          * we do not need to free the memory here, as the
 783                          * buf is deleted by the caller.
 784                          */
 785                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 786                         if (!obj)
 787                                 die(_("invalid %s"), typename(type));
 788                         if (do_fsck_object &&
 789                             fsck_object(obj, buf, size, 1,
 790                                     fsck_error_function))
 791                                 die(_("Error in object"));
 792                         if (fsck_walk(obj, mark_link, NULL))
 793                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 794
 795                         if (obj->type == OBJ_TREE) {
 796                                 struct tree *item = (struct tree *) obj;
 797                                 item->buffer = NULL;
 798                                 obj->parsed = 0;
 799                         }
 800                         if (obj->type == OBJ_COMMIT) {
 801                                 struct commit *commit = (struct commit *) obj;
 802                                 if (detach_commit_buffer(commit, NULL) != data)
 803                                         die("BUG: parse_object_buffer transmogrified our buffer");
 804                         }
 805                         obj->flags |= FLAG_CHECKED;
 806                 }
 807                 read_unlock();
 808         }
 809
 810         free(new_data);
 811 }
 812
 813 /*
 814  * This function is part of find_unresolved_deltas(). There are two
 815  * walkers going in the opposite ways.
 816  *
 817  * The first one in find_unresolved_deltas() traverses down from
 818  * parent node to children, deflating nodes along the way. However,
 819  * memory for deflated nodes is limited by delta_base_cache_limit, so
 820  * at some point parent node's deflated content may be freed.
 821  *
 822  * The second walker is this function, which goes from current node up
 823  * to top parent if necessary to deflate the node. In normal
 824  * situation, its parent node would be already deflated, so it just
 825  * needs to apply delta.
 826  *
 827  * In the worst case scenario, parent node is no longer deflated because
 828  * we're running out of delta_base_cache_limit; we need to re-deflate
 829  * parents, possibly up to the top base.
 830  *
 831  * All deflated objects here are subject to be freed if we exceed
 832  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 833  * just need to make sure the last node is not freed.
 834  */
 835 static void *get_base_data(struct base_data *c)
 836 {
 837         if (!c->data) {
 838                 struct object_entry *obj = c->obj;
 839                 struct base_data **delta = NULL;
 840                 int delta_nr = 0, delta_alloc = 0;
 841
 842                 while (is_delta_type(c->obj->type) && !c->data) {
 843                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 844                         delta[delta_nr++] = c;
 845                         c = c->base;
 846                 }
 847                 if (!delta_nr) {
 848                         c->data = get_data_from_pack(obj);
 849                         c->size = obj->size;
 850                         get_thread_data()->base_cache_used += c->size;
 851                         prune_base_data(c);
 852                 }
 853                 for (; delta_nr > 0; delta_nr--) {
 854                         void *base, *raw;
 855                         c = delta[delta_nr - 1];
 856                         obj = c->obj;
 857                         base = get_base_data(c->base);
 858                         raw = get_data_from_pack(obj);
 859                         c->data = patch_delta(
 860                                 base, c->base->size,
 861                                 raw, obj->size,
 862                                 &c->size);
 863                         free(raw);
 864                         if (!c->data)
 865                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 866                         get_thread_data()->base_cache_used += c->size;
 867                         prune_base_data(c);
 868                 }
 869                 free(delta);
 870         }
 871         return c->data;
 872 }
 873
 874 static void resolve_delta(struct object_entry *delta_obj,
 875                           struct base_data *base, struct base_data *result)
 876 {
 877         void *base_data, *delta_data;
 878
 879         if (show_stat) {
 880                 int i = delta_obj - objects;
 881                 int j = base->obj - objects;
 882                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 883                 deepest_delta_lock();
 884                 if (deepest_delta < obj_stat[i].delta_depth)
 885                         deepest_delta = obj_stat[i].delta_depth;
 886                 deepest_delta_unlock();
 887                 obj_stat[i].base_object_no = j;
 888         }
 889         delta_data = get_data_from_pack(delta_obj);
 890         base_data = get_base_data(base);
 891         result->obj = delta_obj;
 892         result->data = patch_delta(base_data, base->size,
 893                                    delta_data, delta_obj->size, &result->size);
 894         free(delta_data);
 895         if (!result->data)
 896                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 897         hash_sha1_file(result->data, result->size,
 898                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 899         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 900                     delta_obj->idx.sha1);
 901         counter_lock();
 902         nr_resolved_deltas++;
 903         counter_unlock();
 904 }
 905
 906 /*
 907  * Standard boolean compare-and-swap: atomically check whether "*type" is
 908  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 909  * and return false.
 910  */
 911 static int compare_and_swap_type(signed char *type,
 912                                  enum object_type want,
 913                                  enum object_type set)
 914 {
 915         enum object_type old;
 916
 917         type_cas_lock();
 918         old = *type;
 919         if (old == want)
 920                 *type = set;
 921         type_cas_unlock();
 922
 923         return old == want;
 924 }
 925
 926 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 927                                                   struct base_data *prev_base)
 928 {
 929         if (base->ref_last == -1 && base->ofs_last == -1) {
 930                 union delta_base base_spec;
 931
 932                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 933                 find_delta_children(&base_spec,
 934                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 935
 936                 memset(&base_spec, 0, sizeof(base_spec));
 937                 base_spec.offset = base->obj->idx.offset;
 938                 find_delta_children(&base_spec,
 939                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 940
 941                 if (base->ref_last == -1 && base->ofs_last == -1) {
 942                         free(base->data);
 943                         return NULL;
 944                 }
 945
 946                 link_base_data(prev_base, base);
 947         }
 948
 949         if (base->ref_first <= base->ref_last) {
 950                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 951                 struct base_data *result = alloc_base_data();
 952
 953                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
 954                                            base->obj->real_type))
 955                         die("BUG: child->real_type != OBJ_REF_DELTA");
 956
 957                 resolve_delta(child, base, result);
 958                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 959                         free_base_data(base);
 960
 961                 base->ref_first++;
 962                 return result;
 963         }
 964
 965         if (base->ofs_first <= base->ofs_last) {
 966                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 967                 struct base_data *result = alloc_base_data();
 968
 969                 assert(child->real_type == OBJ_OFS_DELTA);
 970                 child->real_type = base->obj->real_type;
 971                 resolve_delta(child, base, result);
 972                 if (base->ofs_first == base->ofs_last)
 973                         free_base_data(base);
 974
 975                 base->ofs_first++;
 976                 return result;
 977         }
 978
 979         unlink_base_data(base);
 980         return NULL;
 981 }
 982
 983 static void find_unresolved_deltas(struct base_data *base)
 984 {
 985         struct base_data *new_base, *prev_base = NULL;
 986         for (;;) {
 987                 new_base = find_unresolved_deltas_1(base, prev_base);
 988
 989                 if (new_base) {
 990                         prev_base = base;
 991                         base = new_base;
 992                 } else {
 993                         free(base);
 994                         base = prev_base;
 995                         if (!base)
 996                                 return;
 997                         prev_base = base->base;
 998                 }
 999         }
1000 }
1001
1002 static int compare_delta_entry(const void *a, const void *b)
1003 {
1004         const struct delta_entry *delta_a = a;
1005         const struct delta_entry *delta_b = b;
1006
1007         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
1008         return compare_delta_bases(&delta_a->base, &delta_b->base,
1009                                    objects[delta_a->obj_no].type,
1010                                    objects[delta_b->obj_no].type);
1011 }
1012
1013 static void resolve_base(struct object_entry *obj)
1014 {
1015         struct base_data *base_obj = alloc_base_data();
1016         base_obj->obj = obj;
1017         base_obj->data = NULL;
1018         find_unresolved_deltas(base_obj);
1019 }
1020
1021 #ifndef NO_PTHREADS
1022 static void *threaded_second_pass(void *data)
1023 {
1024         set_thread_data(data);
1025         for (;;) {
1026                 int i;
1027                 counter_lock();
1028                 display_progress(progress, nr_resolved_deltas);
1029                 counter_unlock();
1030                 work_lock();
1031                 while (nr_dispatched < nr_objects &&
1032                        is_delta_type(objects[nr_dispatched].type))
1033                         nr_dispatched++;
1034                 if (nr_dispatched >= nr_objects) {
1035                         work_unlock();
1036                         break;
1037                 }
1038                 i = nr_dispatched++;
1039                 work_unlock();
1040
1041                 resolve_base(&objects[i]);
1042         }
1043         return NULL;
1044 }
1045 #endif
1046
1047 /*
1048  * First pass:
1049  * - find locations of all objects;
1050  * - calculate SHA1 of all non-delta objects;
1051  * - remember base (SHA1 or offset) for all deltas.
1052  */
1053 static void parse_pack_objects(unsigned char *sha1)
1054 {
1055         int i, nr_delays = 0;
1056         struct delta_entry *delta = deltas;
1057         struct stat st;
1058
1059         if (verbose)
1060                 progress = start_progress(
1061                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1062                                 nr_objects);
1063         for (i = 0; i < nr_objects; i++) {
1064                 struct object_entry *obj = &objects[i];
1065                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1066                 obj->real_type = obj->type;
1067                 if (is_delta_type(obj->type)) {
1068                         nr_deltas++;
1069                         delta->obj_no = i;
1070                         delta++;
1071                 } else if (!data) {
1072                         /* large blobs, check later */
1073                         obj->real_type = OBJ_BAD;
1074                         nr_delays++;
1075                 } else
1076                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1077                 free(data);
1078                 display_progress(progress, i+1);
1079         }
1080         objects[i].idx.offset = consumed_bytes;
1081         stop_progress(&progress);
1082
1083         /* Check pack integrity */
1084         flush();
1085         git_SHA1_Final(sha1, &input_ctx);
1086         if (hashcmp(fill(20), sha1))
1087                 die(_("pack is corrupted (SHA1 mismatch)"));
1088         use(20);
1089
1090         /* If input_fd is a file, we should have reached its end now. */
1091         if (fstat(input_fd, &st))
1092                 die_errno(_("cannot fstat packfile"));
1093         if (S_ISREG(st.st_mode) &&
1094                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1095                 die(_("pack has junk at the end"));
1096
1097         for (i = 0; i < nr_objects; i++) {
1098                 struct object_entry *obj = &objects[i];
1099                 if (obj->real_type != OBJ_BAD)
1100                         continue;
1101                 obj->real_type = obj->type;
1102                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1103                 nr_delays--;
1104         }
1105         if (nr_delays)
1106                 die(_("confusion beyond insanity in parse_pack_objects()"));
1107 }
1108
1109 /*
1110  * Second pass:
1111  * - for all non-delta objects, look if it is used as a base for
1112  *   deltas;
1113  * - if used as a base, uncompress the object and apply all deltas,
1114  *   recursively checking if the resulting object is used as a base
1115  *   for some more deltas.
1116  */
1117 static void resolve_deltas(void)
1118 {
1119         int i;
1120
1121         if (!nr_deltas)
1122                 return;
1123
1124         /* Sort deltas by base SHA1/offset for fast searching */
1125         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1126               compare_delta_entry);
1127
1128         if (verbose)
1129                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1130
1131 #ifndef NO_PTHREADS
1132         nr_dispatched = 0;
1133         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1134                 init_thread();
1135                 for (i = 0; i < nr_threads; i++) {
1136                         int ret = pthread_create(&thread_data[i].thread, NULL,
1137                                                  threaded_second_pass, thread_data + i);
1138                         if (ret)
1139                                 die(_("unable to create thread: %s"),
1140                                     strerror(ret));
1141                 }
1142                 for (i = 0; i < nr_threads; i++)
1143                         pthread_join(thread_data[i].thread, NULL);
1144                 cleanup_thread();
1145                 return;
1146         }
1147 #endif
1148
1149         for (i = 0; i < nr_objects; i++) {
1150                 struct object_entry *obj = &objects[i];
1151
1152                 if (is_delta_type(obj->type))
1153                         continue;
1154                 resolve_base(obj);
1155                 display_progress(progress, nr_resolved_deltas);
1156         }
1157 }
1158
1159 /*
1160  * Third pass:
1161  * - append objects to convert thin pack to full pack if required
1162  * - write the final 20-byte SHA-1
1163  */
1164 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1165 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1166 {
1167         if (nr_deltas == nr_resolved_deltas) {
1168                 stop_progress(&progress);
1169                 /* Flush remaining pack final 20-byte SHA1. */
1170                 flush();
1171                 return;
1172         }
1173
1174         if (fix_thin_pack) {
1175                 struct sha1file *f;
1176                 unsigned char read_sha1[20], tail_sha1[20];
1177                 struct strbuf msg = STRBUF_INIT;
1178                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1179                 int nr_objects_initial = nr_objects;
1180                 if (nr_unresolved <= 0)
1181                         die(_("confusion beyond insanity"));
1182                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1183                 memset(objects + nr_objects + 1, 0,
1184                        nr_unresolved * sizeof(*objects));
1185                 f = sha1fd(output_fd, curr_pack);
1186                 fix_unresolved_deltas(f, nr_unresolved);
1187                 strbuf_addf(&msg, _("completed with %d local objects"),
1188                             nr_objects - nr_objects_initial);
1189                 stop_progress_msg(&progress, msg.buf);
1190                 strbuf_release(&msg);
1191                 sha1close(f, tail_sha1, 0);
1192                 hashcpy(read_sha1, pack_sha1);
1193                 fixup_pack_header_footer(output_fd, pack_sha1,
1194                                          curr_pack, nr_objects,
1195                                          read_sha1, consumed_bytes-20);
1196                 if (hashcmp(read_sha1, tail_sha1) != 0)
1197                         die(_("Unexpected tail checksum for %s "
1198                               "(disk corruption?)"), curr_pack);
1199         }
1200         if (nr_deltas != nr_resolved_deltas)
1201                 die(Q_("pack has %d unresolved delta",
1202                        "pack has %d unresolved deltas",
1203                        nr_deltas - nr_resolved_deltas),
1204                     nr_deltas - nr_resolved_deltas);
1205 }
1206
1207 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1208 {
1209         git_zstream stream;
1210         int status;
1211         unsigned char outbuf[4096];
1212
1213         memset(&stream, 0, sizeof(stream));
1214         git_deflate_init(&stream, zlib_compression_level);
1215         stream.next_in = in;
1216         stream.avail_in = size;
1217
1218         do {
1219                 stream.next_out = outbuf;
1220                 stream.avail_out = sizeof(outbuf);
1221                 status = git_deflate(&stream, Z_FINISH);
1222                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1223         } while (status == Z_OK);
1224
1225         if (status != Z_STREAM_END)
1226                 die(_("unable to deflate appended object (%d)"), status);
1227         size = stream.total_out;
1228         git_deflate_end(&stream);
1229         return size;
1230 }
1231
1232 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1233                                const unsigned char *sha1, void *buf,
1234                                unsigned long size, enum object_type type)
1235 {
1236         struct object_entry *obj = &objects[nr_objects++];
1237         unsigned char header[10];
1238         unsigned long s = size;
1239         int n = 0;
1240         unsigned char c = (type << 4) | (s & 15);
1241         s >>= 4;
1242         while (s) {
1243                 header[n++] = c | 0x80;
1244                 c = s & 0x7f;
1245                 s >>= 7;
1246         }
1247         header[n++] = c;
1248         crc32_begin(f);
1249         sha1write(f, header, n);
1250         obj[0].size = size;
1251         obj[0].hdr_size = n;
1252         obj[0].type = type;
1253         obj[0].real_type = type;
1254         obj[1].idx.offset = obj[0].idx.offset + n;
1255         obj[1].idx.offset += write_compressed(f, buf, size);
1256         obj[0].idx.crc32 = crc32_end(f);
1257         sha1flush(f);
1258         hashcpy(obj->idx.sha1, sha1);
1259         return obj;
1260 }
1261
1262 static int delta_pos_compare(const void *_a, const void *_b)
1263 {
1264         struct delta_entry *a = *(struct delta_entry **)_a;
1265         struct delta_entry *b = *(struct delta_entry **)_b;
1266         return a->obj_no - b->obj_no;
1267 }
1268
1269 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1270 {
1271         struct delta_entry **sorted_by_pos;
1272         int i, n = 0;
1273
1274         /*
1275          * Since many unresolved deltas may well be themselves base objects
1276          * for more unresolved deltas, we really want to include the
1277          * smallest number of base objects that would cover as much delta
1278          * as possible by picking the
1279          * trunc deltas first, allowing for other deltas to resolve without
1280          * additional base objects.  Since most base objects are to be found
1281          * before deltas depending on them, a good heuristic is to start
1282          * resolving deltas in the same order as their position in the pack.
1283          */
1284         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1285         for (i = 0; i < nr_deltas; i++) {
1286                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1287                         continue;
1288                 sorted_by_pos[n++] = &deltas[i];
1289         }
1290         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1291
1292         for (i = 0; i < n; i++) {
1293                 struct delta_entry *d = sorted_by_pos[i];
1294                 enum object_type type;
1295                 struct base_data *base_obj = alloc_base_data();
1296
1297                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1298                         continue;
1299                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1300                 if (!base_obj->data)
1301                         continue;
1302
1303                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1304                                 base_obj->size, typename(type)))
1305                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1306                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1307                                         base_obj->data, base_obj->size, type);
1308                 find_unresolved_deltas(base_obj);
1309                 display_progress(progress, nr_resolved_deltas);
1310         }
1311         free(sorted_by_pos);
1312 }
1313
1314 static void final(const char *final_pack_name, const char *curr_pack_name,
1315                   const char *final_index_name, const char *curr_index_name,
1316                   const char *keep_name, const char *keep_msg,
1317                   unsigned char *sha1)
1318 {
1319         const char *report = "pack";
1320         char name[PATH_MAX];
1321         int err;
1322
1323         if (!from_stdin) {
1324                 close(input_fd);
1325         } else {
1326                 fsync_or_die(output_fd, curr_pack_name);
1327                 err = close(output_fd);
1328                 if (err)
1329                         die_errno(_("error while closing pack file"));
1330         }
1331
1332         if (keep_msg) {
1333                 int keep_fd, keep_msg_len = strlen(keep_msg);
1334
1335                 if (!keep_name)
1336                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1337                 else
1338                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1339
1340                 if (keep_fd < 0) {
1341                         if (errno != EEXIST)
1342                                 die_errno(_("cannot write keep file '%s'"),
1343                                           keep_name ? keep_name : name);
1344                 } else {
1345                         if (keep_msg_len > 0) {
1346                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1347                                 write_or_die(keep_fd, "\n", 1);
1348                         }
1349                         if (close(keep_fd) != 0)
1350                                 die_errno(_("cannot close written keep file '%s'"),
1351                                           keep_name ? keep_name : name);
1352                         report = "keep";
1353                 }
1354         }
1355
1356         if (final_pack_name != curr_pack_name) {
1357                 if (!final_pack_name) {
1358                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1359                                  get_object_directory(), sha1_to_hex(sha1));
1360                         final_pack_name = name;
1361                 }
1362                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1363                         die(_("cannot store pack file"));
1364         } else if (from_stdin)
1365                 chmod(final_pack_name, 0444);
1366
1367         if (final_index_name != curr_index_name) {
1368                 if (!final_index_name) {
1369                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1370                                  get_object_directory(), sha1_to_hex(sha1));
1371                         final_index_name = name;
1372                 }
1373                 if (move_temp_to_file(curr_index_name, final_index_name))
1374                         die(_("cannot store index file"));
1375         } else
1376                 chmod(final_index_name, 0444);
1377
1378         if (!from_stdin) {
1379                 printf("%s\n", sha1_to_hex(sha1));
1380         } else {
1381                 char buf[48];
1382                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1383                                    report, sha1_to_hex(sha1));
1384                 write_or_die(1, buf, len);
1385
1386                 /*
1387                  * Let's just mimic git-unpack-objects here and write
1388                  * the last part of the input buffer to stdout.
1389                  */
1390                 while (input_len) {
1391                         err = xwrite(1, input_buffer + input_offset, input_len);
1392                         if (err <= 0)
1393                                 break;
1394                         input_len -= err;
1395                         input_offset += err;
1396                 }
1397         }
1398 }
1399
1400 static int git_index_pack_config(const char *k, const char *v, void *cb)
1401 {
1402         struct pack_idx_option *opts = cb;
1403
1404         if (!strcmp(k, "pack.indexversion")) {
1405                 opts->version = git_config_int(k, v);
1406                 if (opts->version > 2)
1407                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1408                 return 0;
1409         }
1410         if (!strcmp(k, "pack.threads")) {
1411                 nr_threads = git_config_int(k, v);
1412                 if (nr_threads < 0)
1413                         die(_("invalid number of threads specified (%d)"),
1414                             nr_threads);
1415 #ifdef NO_PTHREADS
1416                 if (nr_threads != 1)
1417                         warning(_("no threads support, ignoring %s"), k);
1418                 nr_threads = 1;
1419 #endif
1420                 return 0;
1421         }
1422         return git_default_config(k, v, cb);
1423 }
1424
1425 static int cmp_uint32(const void *a_, const void *b_)
1426 {
1427         uint32_t a = *((uint32_t *)a_);
1428         uint32_t b = *((uint32_t *)b_);
1429
1430         return (a < b) ? -1 : (a != b);
1431 }
1432
1433 static void read_v2_anomalous_offsets(struct packed_git *p,
1434                                       struct pack_idx_option *opts)
1435 {
1436         const uint32_t *idx1, *idx2;
1437         uint32_t i;
1438
1439         /* The address of the 4-byte offset table */
1440         idx1 = (((const uint32_t *)p->index_data)
1441                 + 2 /* 8-byte header */
1442                 + 256 /* fan out */
1443                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1444                 + p->num_objects /* CRC32 table */
1445                 );
1446
1447         /* The address of the 8-byte offset table */
1448         idx2 = idx1 + p->num_objects;
1449
1450         for (i = 0; i < p->num_objects; i++) {
1451                 uint32_t off = ntohl(idx1[i]);
1452                 if (!(off & 0x80000000))
1453                         continue;
1454                 off = off & 0x7fffffff;
1455                 if (idx2[off * 2])
1456                         continue;
1457                 /*
1458                  * The real offset is ntohl(idx2[off * 2]) in high 4
1459                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1460                  * octets.  But idx2[off * 2] is Zero!!!
1461                  */
1462                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1463                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1464         }
1465
1466         if (1 < opts->anomaly_nr)
1467                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1468 }
1469
1470 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1471 {
1472         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1473
1474         if (!p)
1475                 die(_("Cannot open existing pack file '%s'"), pack_name);
1476         if (open_pack_index(p))
1477                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1478
1479         /* Read the attributes from the existing idx file */
1480         opts->version = p->index_version;
1481
1482         if (opts->version == 2)
1483                 read_v2_anomalous_offsets(p, opts);
1484
1485         /*
1486          * Get rid of the idx file as we do not need it anymore.
1487          * NEEDSWORK: extract this bit from free_pack_by_name() in
1488          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1489          * know we haven't installed this pack (hence we never have
1490          * read anything from it).
1491          */
1492         close_pack_index(p);
1493         free(p);
1494 }
1495
1496 static void show_pack_info(int stat_only)
1497 {
1498         int i, baseobjects = nr_objects - nr_deltas;
1499         unsigned long *chain_histogram = NULL;
1500
1501         if (deepest_delta)
1502                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1503
1504         for (i = 0; i < nr_objects; i++) {
1505                 struct object_entry *obj = &objects[i];
1506
1507                 if (is_delta_type(obj->type))
1508                         chain_histogram[obj_stat[i].delta_depth - 1]++;
1509                 if (stat_only)
1510                         continue;
1511                 printf("%s %-6s %lu %lu %"PRIuMAX,
1512                        sha1_to_hex(obj->idx.sha1),
1513                        typename(obj->real_type), obj->size,
1514                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1515                        (uintmax_t)obj->idx.offset);
1516                 if (is_delta_type(obj->type)) {
1517                         struct object_entry *bobj = &objects[obj_stat[i].base_object_no];
1518                         printf(" %u %s", obj_stat[i].delta_depth, sha1_to_hex(bobj->idx.sha1));
1519                 }
1520                 putchar('\n');
1521         }
1522
1523         if (baseobjects)
1524                 printf_ln(Q_("non delta: %d object",
1525                              "non delta: %d objects",
1526                              baseobjects),
1527                           baseobjects);
1528         for (i = 0; i < deepest_delta; i++) {
1529                 if (!chain_histogram[i])
1530                         continue;
1531                 printf_ln(Q_("chain length = %d: %lu object",
1532                              "chain length = %d: %lu objects",
1533                              chain_histogram[i]),
1534                           i + 1,
1535                           chain_histogram[i]);
1536         }
1537 }
1538
1539 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1540 {
1541         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1542         const char *curr_index;
1543         const char *index_name = NULL, *pack_name = NULL;
1544         const char *keep_name = NULL, *keep_msg = NULL;
1545         struct strbuf index_name_buf = STRBUF_INIT,
1546                       keep_name_buf = STRBUF_INIT;
1547         struct pack_idx_entry **idx_objects;
1548         struct pack_idx_option opts;
1549         unsigned char pack_sha1[20];
1550         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1551
1552         if (argc == 2 && !strcmp(argv[1], "-h"))
1553                 usage(index_pack_usage);
1554
1555         check_replace_refs = 0;
1556
1557         reset_pack_idx_option(&opts);
1558         git_config(git_index_pack_config, &opts);
1559         if (prefix && chdir(prefix))
1560                 die(_("Cannot come back to cwd"));
1561
1562         for (i = 1; i < argc; i++) {
1563                 const char *arg = argv[i];
1564
1565                 if (*arg == '-') {
1566                         if (!strcmp(arg, "--stdin")) {
1567                                 from_stdin = 1;
1568                         } else if (!strcmp(arg, "--fix-thin")) {
1569                                 fix_thin_pack = 1;
1570                         } else if (!strcmp(arg, "--strict")) {
1571                                 strict = 1;
1572                                 do_fsck_object = 1;
1573                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1574                                 strict = 1;
1575                                 check_self_contained_and_connected = 1;
1576                         } else if (!strcmp(arg, "--verify")) {
1577                                 verify = 1;
1578                         } else if (!strcmp(arg, "--verify-stat")) {
1579                                 verify = 1;
1580                                 show_stat = 1;
1581                         } else if (!strcmp(arg, "--verify-stat-only")) {
1582                                 verify = 1;
1583                                 show_stat = 1;
1584                                 stat_only = 1;
1585                         } else if (!strcmp(arg, "--keep")) {
1586                                 keep_msg = "";
1587                         } else if (starts_with(arg, "--keep=")) {
1588                                 keep_msg = arg + 7;
1589                         } else if (starts_with(arg, "--threads=")) {
1590                                 char *end;
1591                                 nr_threads = strtoul(arg+10, &end, 0);
1592                                 if (!arg[10] || *end || nr_threads < 0)
1593                                         usage(index_pack_usage);
1594 #ifdef NO_PTHREADS
1595                                 if (nr_threads != 1)
1596                                         warning(_("no threads support, "
1597                                                   "ignoring %s"), arg);
1598                                 nr_threads = 1;
1599 #endif
1600                         } else if (starts_with(arg, "--pack_header=")) {
1601                                 struct pack_header *hdr;
1602                                 char *c;
1603
1604                                 hdr = (struct pack_header *)input_buffer;
1605                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1606                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1607                                 if (*c != ',')
1608                                         die(_("bad %s"), arg);
1609                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1610                                 if (*c)
1611                                         die(_("bad %s"), arg);
1612                                 input_len = sizeof(*hdr);
1613                         } else if (!strcmp(arg, "-v")) {
1614                                 verbose = 1;
1615                         } else if (!strcmp(arg, "-o")) {
1616                                 if (index_name || (i+1) >= argc)
1617                                         usage(index_pack_usage);
1618                                 index_name = argv[++i];
1619                         } else if (starts_with(arg, "--index-version=")) {
1620                                 char *c;
1621                                 opts.version = strtoul(arg + 16, &c, 10);
1622                                 if (opts.version > 2)
1623                                         die(_("bad %s"), arg);
1624                                 if (*c == ',')
1625                                         opts.off32_limit = strtoul(c+1, &c, 0);
1626                                 if (*c || opts.off32_limit & 0x80000000)
1627                                         die(_("bad %s"), arg);
1628                         } else
1629                                 usage(index_pack_usage);
1630                         continue;
1631                 }
1632
1633                 if (pack_name)
1634                         usage(index_pack_usage);
1635                 pack_name = arg;
1636         }
1637
1638         if (!pack_name && !from_stdin)
1639                 usage(index_pack_usage);
1640         if (fix_thin_pack && !from_stdin)
1641                 die(_("--fix-thin cannot be used without --stdin"));
1642         if (!index_name && pack_name) {
1643                 size_t len;
1644                 if (!strip_suffix(pack_name, ".pack", &len))
1645                         die(_("packfile name '%s' does not end with '.pack'"),
1646                             pack_name);
1647                 strbuf_add(&index_name_buf, pack_name, len);
1648                 strbuf_addstr(&index_name_buf, ".idx");
1649                 index_name = index_name_buf.buf;
1650         }
1651         if (keep_msg && !keep_name && pack_name) {
1652                 size_t len;
1653                 if (!strip_suffix(pack_name, ".pack", &len))
1654                         die(_("packfile name '%s' does not end with '.pack'"),
1655                             pack_name);
1656                 strbuf_add(&keep_name_buf, pack_name, len);
1657                 strbuf_addstr(&keep_name_buf, ".idx");
1658                 keep_name = keep_name_buf.buf;
1659         }
1660         if (verify) {
1661                 if (!index_name)
1662                         die(_("--verify with no packfile name given"));
1663                 read_idx_option(&opts, index_name);
1664                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1665         }
1666         if (strict)
1667                 opts.flags |= WRITE_IDX_STRICT;
1668
1669 #ifndef NO_PTHREADS
1670         if (!nr_threads) {
1671                 nr_threads = online_cpus();
1672                 /* An experiment showed that more threads does not mean faster */
1673                 if (nr_threads > 3)
1674                         nr_threads = 3;
1675         }
1676 #endif
1677
1678         curr_pack = open_pack_file(pack_name);
1679         parse_pack_header();
1680         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1681         if (show_stat)
1682                 obj_stat = xcalloc(nr_objects + 1, sizeof(struct object_stat));
1683         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1684         parse_pack_objects(pack_sha1);
1685         resolve_deltas();
1686         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1687         free(deltas);
1688         if (strict)
1689                 foreign_nr = check_objects();
1690
1691         if (show_stat)
1692                 show_pack_info(stat_only);
1693
1694         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1695         for (i = 0; i < nr_objects; i++)
1696                 idx_objects[i] = &objects[i].idx;
1697         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1698         free(idx_objects);
1699
1700         if (!verify)
1701                 final(pack_name, curr_pack,
1702                       index_name, curr_index,
1703                       keep_name, keep_msg,
1704                       pack_sha1);
1705         else
1706                 close(input_fd);
1707         free(objects);
1708         strbuf_release(&index_name_buf);
1709         strbuf_release(&keep_name_buf);
1710         if (pack_name == NULL)
1711                 free((void *) curr_pack);
1712         if (index_name == NULL)
1713                 free((void *) curr_index);
1714
1715         /*
1716          * Let the caller know this pack is not self contained
1717          */
1718         if (check_self_contained_and_connected && foreign_nr)
1719                 return 1;
1720
1721         return 0;
1722 }