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 unsigned deepest_delta;
  91 static git_SHA_CTX input_ctx;
  92 static uint32_t input_crc32;
  93 static int input_fd, output_fd;
  94 static const char *curr_pack;
  95
  96 #ifndef NO_PTHREADS
  97
  98 static struct thread_local *thread_data;
  99 static int nr_dispatched;
 100 static int threads_active;
 101
 102 static pthread_mutex_t read_mutex;
 103 #define read_lock()             lock_mutex(&read_mutex)
 104 #define read_unlock()           unlock_mutex(&read_mutex)
 105
 106 static pthread_mutex_t counter_mutex;
 107 #define counter_lock()          lock_mutex(&counter_mutex)
 108 #define counter_unlock()        unlock_mutex(&counter_mutex)
 109
 110 static pthread_mutex_t work_mutex;
 111 #define work_lock()             lock_mutex(&work_mutex)
 112 #define work_unlock()           unlock_mutex(&work_mutex)
 113
 114 static pthread_mutex_t deepest_delta_mutex;
 115 #define deepest_delta_lock()    lock_mutex(&deepest_delta_mutex)
 116 #define deepest_delta_unlock()  unlock_mutex(&deepest_delta_mutex)
 117
 118 static pthread_mutex_t type_cas_mutex;
 119 #define type_cas_lock()         lock_mutex(&type_cas_mutex)
 120 #define type_cas_unlock()       unlock_mutex(&type_cas_mutex)
 121
 122 static pthread_key_t key;
 123
 124 static inline void lock_mutex(pthread_mutex_t *mutex)
 125 {
 126         if (threads_active)
 127                 pthread_mutex_lock(mutex);
 128 }
 129
 130 static inline void unlock_mutex(pthread_mutex_t *mutex)
 131 {
 132         if (threads_active)
 133                 pthread_mutex_unlock(mutex);
 134 }
 135
 136 /*
 137  * Mutex and conditional variable can't be statically-initialized on Windows.
 138  */
 139 static void init_thread(void)
 140 {
 141         int i;
 142         init_recursive_mutex(&read_mutex);
 143         pthread_mutex_init(&counter_mutex, NULL);
 144         pthread_mutex_init(&work_mutex, NULL);
 145         pthread_mutex_init(&type_cas_mutex, NULL);
 146         if (show_stat)
 147                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 148         pthread_key_create(&key, NULL);
 149         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 150         for (i = 0; i < nr_threads; i++) {
 151                 thread_data[i].pack_fd = open(curr_pack, O_RDONLY);
 152                 if (thread_data[i].pack_fd == -1)
 153                         die_errno(_("unable to open %s"), curr_pack);
 154         }
 155
 156         threads_active = 1;
 157 }
 158
 159 static void cleanup_thread(void)
 160 {
 161         int i;
 162         if (!threads_active)
 163                 return;
 164         threads_active = 0;
 165         pthread_mutex_destroy(&read_mutex);
 166         pthread_mutex_destroy(&counter_mutex);
 167         pthread_mutex_destroy(&work_mutex);
 168         pthread_mutex_destroy(&type_cas_mutex);
 169         if (show_stat)
 170                 pthread_mutex_destroy(&deepest_delta_mutex);
 171         for (i = 0; i < nr_threads; i++)
 172                 close(thread_data[i].pack_fd);
 173         pthread_key_delete(key);
 174         free(thread_data);
 175 }
 176
 177 #else
 178
 179 #define read_lock()
 180 #define read_unlock()
 181
 182 #define counter_lock()
 183 #define counter_unlock()
 184
 185 #define work_lock()
 186 #define work_unlock()
 187
 188 #define deepest_delta_lock()
 189 #define deepest_delta_unlock()
 190
 191 #define type_cas_lock()
 192 #define type_cas_unlock()
 193
 194 #endif
 195
 196
 197 static int mark_link(struct object *obj, int type, void *data, struct fsck_options *options)
 198 {
 199         if (!obj)
 200                 return -1;
 201
 202         if (type != OBJ_ANY && obj->type != type)
 203                 die(_("object type mismatch at %s"), oid_to_hex(&obj->oid));
 204
 205         obj->flags |= FLAG_LINK;
 206         return 0;
 207 }
 208
 209 /* The content of each linked object must have been checked
 210    or it must be already present in the object database */
 211 static unsigned check_object(struct object *obj)
 212 {
 213         if (!obj)
 214                 return 0;
 215
 216         if (!(obj->flags & FLAG_LINK))
 217                 return 0;
 218
 219         if (!(obj->flags & FLAG_CHECKED)) {
 220                 unsigned long size;
 221                 int type = sha1_object_info(obj->oid.hash, &size);
 222                 if (type <= 0)
 223                         die(_("did not receive expected object %s"),
 224                               oid_to_hex(&obj->oid));
 225                 if (type != obj->type)
 226                         die(_("object %s: expected type %s, found %s"),
 227                             oid_to_hex(&obj->oid),
 228                             typename(obj->type), typename(type));
 229                 obj->flags |= FLAG_CHECKED;
 230                 return 1;
 231         }
 232
 233         return 0;
 234 }
 235
 236 static unsigned check_objects(void)
 237 {
 238         unsigned i, max, foreign_nr = 0;
 239
 240         max = get_max_object_index();
 241         for (i = 0; i < max; i++)
 242                 foreign_nr += check_object(get_indexed_object(i));
 243         return foreign_nr;
 244 }
 245
 246
 247 /* Discard current buffer used content. */
 248 static void flush(void)
 249 {
 250         if (input_offset) {
 251                 if (output_fd >= 0)
 252                         write_or_die(output_fd, input_buffer, input_offset);
 253                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 254                 memmove(input_buffer, input_buffer + input_offset, input_len);
 255                 input_offset = 0;
 256         }
 257 }
 258
 259 /*
 260  * Make sure at least "min" bytes are available in the buffer, and
 261  * return the pointer to the buffer.
 262  */
 263 static void *fill(int min)
 264 {
 265         if (min <= input_len)
 266                 return input_buffer + input_offset;
 267         if (min > sizeof(input_buffer))
 268                 die(Q_("cannot fill %d byte",
 269                        "cannot fill %d bytes",
 270                        min),
 271                     min);
 272         flush();
 273         do {
 274                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 275                                 sizeof(input_buffer) - input_len);
 276                 if (ret <= 0) {
 277                         if (!ret)
 278                                 die(_("early EOF"));
 279                         die_errno(_("read error on input"));
 280                 }
 281                 input_len += ret;
 282                 if (from_stdin)
 283                         display_throughput(progress, consumed_bytes + input_len);
 284         } while (input_len < min);
 285         return input_buffer;
 286 }
 287
 288 static void use(int bytes)
 289 {
 290         if (bytes > input_len)
 291                 die(_("used more bytes than were available"));
 292         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 293         input_len -= bytes;
 294         input_offset += bytes;
 295
 296         /* make sure off_t is sufficiently large not to wrap */
 297         if (signed_add_overflows(consumed_bytes, bytes))
 298                 die(_("pack too large for current definition of off_t"));
 299         consumed_bytes += bytes;
 300 }
 301
 302 static const char *open_pack_file(const char *pack_name)
 303 {
 304         if (from_stdin) {
 305                 input_fd = 0;
 306                 if (!pack_name) {
 307                         static char tmp_file[PATH_MAX];
 308                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 309                                                 "pack/tmp_pack_XXXXXX");
 310                         pack_name = xstrdup(tmp_file);
 311                 } else
 312                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 313                 if (output_fd < 0)
 314                         die_errno(_("unable to create '%s'"), pack_name);
 315                 nothread_data.pack_fd = output_fd;
 316         } else {
 317                 input_fd = open(pack_name, O_RDONLY);
 318                 if (input_fd < 0)
 319                         die_errno(_("cannot open packfile '%s'"), pack_name);
 320                 output_fd = -1;
 321                 nothread_data.pack_fd = input_fd;
 322         }
 323         git_SHA1_Init(&input_ctx);
 324         return pack_name;
 325 }
 326
 327 static void parse_pack_header(void)
 328 {
 329         struct pack_header *hdr = fill(sizeof(struct pack_header));
 330
 331         /* Header consistency check */
 332         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 333                 die(_("pack signature mismatch"));
 334         if (!pack_version_ok(hdr->hdr_version))
 335                 die(_("pack version %"PRIu32" unsupported"),
 336                         ntohl(hdr->hdr_version));
 337
 338         nr_objects = ntohl(hdr->hdr_entries);
 339         use(sizeof(struct pack_header));
 340 }
 341
 342 static NORETURN void bad_object(off_t offset, const char *format,
 343                        ...) __attribute__((format (printf, 2, 3)));
 344
 345 static NORETURN void bad_object(off_t offset, const char *format, ...)
 346 {
 347         va_list params;
 348         char buf[1024];
 349
 350         va_start(params, format);
 351         vsnprintf(buf, sizeof(buf), format, params);
 352         va_end(params);
 353         die(_("pack has bad object at offset %"PRIuMAX": %s"),
 354             (uintmax_t)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(off_t 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 = xsnprintf(hdr, sizeof(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                               off_t *ofs_offset,
 484                               unsigned char *ref_sha1,
 485                               unsigned char *sha1)
 486 {
 487         unsigned char *p;
 488         unsigned long size, c;
 489         off_t base_offset;
 490         unsigned shift;
 491         void *data;
 492
 493         obj->idx.offset = consumed_bytes;
 494         input_crc32 = crc32(0, NULL, 0);
 495
 496         p = fill(1);
 497         c = *p;
 498         use(1);
 499         obj->type = (c >> 4) & 7;
 500         size = (c & 15);
 501         shift = 4;
 502         while (c & 0x80) {
 503                 p = fill(1);
 504                 c = *p;
 505                 use(1);
 506                 size += (c & 0x7f) << shift;
 507                 shift += 7;
 508         }
 509         obj->size = size;
 510
 511         switch (obj->type) {
 512         case OBJ_REF_DELTA:
 513                 hashcpy(ref_sha1, fill(20));
 514                 use(20);
 515                 break;
 516         case OBJ_OFS_DELTA:
 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                 *ofs_offset = obj->idx.offset - base_offset;
 531                 if (*ofs_offset <= 0 || *ofs_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         off_t 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) ? (int)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) ? (ssize_t)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, %"PRIuMAX" byte missing",
 574                                "premature end of pack file, %"PRIuMAX" bytes missing",
 575                                (unsigned int)len),
 576                             (uintmax_t)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_ofs_delta_bases(off_t offset1, off_t offset2,
 616                                    enum object_type type1,
 617                                    enum object_type type2)
 618 {
 619         int cmp = type1 - type2;
 620         if (cmp)
 621                 return cmp;
 622         return offset1 < offset2 ? -1 :
 623                offset1 > offset2 ?  1 :
 624                0;
 625 }
 626
 627 static int find_ofs_delta(const off_t offset, enum object_type type)
 628 {
 629         int first = 0, last = nr_ofs_deltas;
 630
 631         while (first < last) {
 632                 int next = (first + last) / 2;
 633                 struct ofs_delta_entry *delta = &ofs_deltas[next];
 634                 int cmp;
 635
 636                 cmp = compare_ofs_delta_bases(offset, delta->offset,
 637                                               type, objects[delta->obj_no].type);
 638                 if (!cmp)
 639                         return next;
 640                 if (cmp < 0) {
 641                         last = next;
 642                         continue;
 643                 }
 644                 first = next+1;
 645         }
 646         return -first-1;
 647 }
 648
 649 static void find_ofs_delta_children(off_t offset,
 650                                     int *first_index, int *last_index,
 651                                     enum object_type type)
 652 {
 653         int first = find_ofs_delta(offset, type);
 654         int last = first;
 655         int end = nr_ofs_deltas - 1;
 656
 657         if (first < 0) {
 658                 *first_index = 0;
 659                 *last_index = -1;
 660                 return;
 661         }
 662         while (first > 0 && ofs_deltas[first - 1].offset == offset)
 663                 --first;
 664         while (last < end && ofs_deltas[last + 1].offset == offset)
 665                 ++last;
 666         *first_index = first;
 667         *last_index = last;
 668 }
 669
 670 static int compare_ref_delta_bases(const unsigned char *sha1,
 671                                    const unsigned char *sha2,
 672                                    enum object_type type1,
 673                                    enum object_type type2)
 674 {
 675         int cmp = type1 - type2;
 676         if (cmp)
 677                 return cmp;
 678         return hashcmp(sha1, sha2);
 679 }
 680
 681 static int find_ref_delta(const unsigned char *sha1, enum object_type type)
 682 {
 683         int first = 0, last = nr_ref_deltas;
 684
 685         while (first < last) {
 686                 int next = (first + last) / 2;
 687                 struct ref_delta_entry *delta = &ref_deltas[next];
 688                 int cmp;
 689
 690                 cmp = compare_ref_delta_bases(sha1, delta->sha1,
 691                                               type, objects[delta->obj_no].type);
 692                 if (!cmp)
 693                         return next;
 694                 if (cmp < 0) {
 695                         last = next;
 696                         continue;
 697                 }
 698                 first = next+1;
 699         }
 700         return -first-1;
 701 }
 702
 703 static void find_ref_delta_children(const unsigned char *sha1,
 704                                     int *first_index, int *last_index,
 705                                     enum object_type type)
 706 {
 707         int first = find_ref_delta(sha1, type);
 708         int last = first;
 709         int end = nr_ref_deltas - 1;
 710
 711         if (first < 0) {
 712                 *first_index = 0;
 713                 *last_index = -1;
 714                 return;
 715         }
 716         while (first > 0 && !hashcmp(ref_deltas[first - 1].sha1, sha1))
 717                 --first;
 718         while (last < end && !hashcmp(ref_deltas[last + 1].sha1, sha1))
 719                 ++last;
 720         *first_index = first;
 721         *last_index = last;
 722 }
 723
 724 struct compare_data {
 725         struct object_entry *entry;
 726         struct git_istream *st;
 727         unsigned char *buf;
 728         unsigned long buf_size;
 729 };
 730
 731 static int compare_objects(const unsigned char *buf, unsigned long size,
 732                            void *cb_data)
 733 {
 734         struct compare_data *data = cb_data;
 735
 736         if (data->buf_size < size) {
 737                 free(data->buf);
 738                 data->buf = xmalloc(size);
 739                 data->buf_size = size;
 740         }
 741
 742         while (size) {
 743                 ssize_t len = read_istream(data->st, data->buf, size);
 744                 if (len == 0)
 745                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 746                             sha1_to_hex(data->entry->idx.sha1));
 747                 if (len < 0)
 748                         die(_("unable to read %s"),
 749                             sha1_to_hex(data->entry->idx.sha1));
 750                 if (memcmp(buf, data->buf, len))
 751                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 752                             sha1_to_hex(data->entry->idx.sha1));
 753                 size -= len;
 754                 buf += len;
 755         }
 756         return 0;
 757 }
 758
 759 static int check_collison(struct object_entry *entry)
 760 {
 761         struct compare_data data;
 762         enum object_type type;
 763         unsigned long size;
 764
 765         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 766                 return -1;
 767
 768         memset(&data, 0, sizeof(data));
 769         data.entry = entry;
 770         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 771         if (!data.st)
 772                 return -1;
 773         if (size != entry->size || type != entry->type)
 774                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 775                     sha1_to_hex(entry->idx.sha1));
 776         unpack_data(entry, compare_objects, &data);
 777         close_istream(data.st);
 778         free(data.buf);
 779         return 0;
 780 }
 781
 782 static void sha1_object(const void *data, struct object_entry *obj_entry,
 783                         unsigned long size, enum object_type type,
 784                         const unsigned char *sha1)
 785 {
 786         void *new_data = NULL;
 787         int collision_test_needed;
 788
 789         assert(data || obj_entry);
 790
 791         read_lock();
 792         collision_test_needed = has_sha1_file_with_flags(sha1, HAS_SHA1_QUICK);
 793         read_unlock();
 794
 795         if (collision_test_needed && !data) {
 796                 read_lock();
 797                 if (!check_collison(obj_entry))
 798                         collision_test_needed = 0;
 799                 read_unlock();
 800         }
 801         if (collision_test_needed) {
 802                 void *has_data;
 803                 enum object_type has_type;
 804                 unsigned long has_size;
 805                 read_lock();
 806                 has_type = sha1_object_info(sha1, &has_size);
 807                 if (has_type != type || has_size != size)
 808                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 809                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 810                 read_unlock();
 811                 if (!data)
 812                         data = new_data = get_data_from_pack(obj_entry);
 813                 if (!has_data)
 814                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 815                 if (size != has_size || type != has_type ||
 816                     memcmp(data, has_data, size) != 0)
 817                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 818                 free(has_data);
 819         }
 820
 821         if (strict) {
 822                 read_lock();
 823                 if (type == OBJ_BLOB) {
 824                         struct blob *blob = lookup_blob(sha1);
 825                         if (blob)
 826                                 blob->object.flags |= FLAG_CHECKED;
 827                         else
 828                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 829                 } else {
 830                         struct object *obj;
 831                         int eaten;
 832                         void *buf = (void *) data;
 833
 834                         assert(data && "data can only be NULL for large _blobs_");
 835
 836                         /*
 837                          * we do not need to free the memory here, as the
 838                          * buf is deleted by the caller.
 839                          */
 840                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 841                         if (!obj)
 842                                 die(_("invalid %s"), typename(type));
 843                         if (do_fsck_object &&
 844                             fsck_object(obj, buf, size, &fsck_options))
 845                                 die(_("Error in object"));
 846                         if (fsck_walk(obj, NULL, &fsck_options))
 847                                 die(_("Not all child objects of %s are reachable"), oid_to_hex(&obj->oid));
 848
 849                         if (obj->type == OBJ_TREE) {
 850                                 struct tree *item = (struct tree *) obj;
 851                                 item->buffer = NULL;
 852                                 obj->parsed = 0;
 853                         }
 854                         if (obj->type == OBJ_COMMIT) {
 855                                 struct commit *commit = (struct commit *) obj;
 856                                 if (detach_commit_buffer(commit, NULL) != data)
 857                                         die("BUG: parse_object_buffer transmogrified our buffer");
 858                         }
 859                         obj->flags |= FLAG_CHECKED;
 860                 }
 861                 read_unlock();
 862         }
 863
 864         free(new_data);
 865 }
 866
 867 /*
 868  * This function is part of find_unresolved_deltas(). There are two
 869  * walkers going in the opposite ways.
 870  *
 871  * The first one in find_unresolved_deltas() traverses down from
 872  * parent node to children, deflating nodes along the way. However,
 873  * memory for deflated nodes is limited by delta_base_cache_limit, so
 874  * at some point parent node's deflated content may be freed.
 875  *
 876  * The second walker is this function, which goes from current node up
 877  * to top parent if necessary to deflate the node. In normal
 878  * situation, its parent node would be already deflated, so it just
 879  * needs to apply delta.
 880  *
 881  * In the worst case scenario, parent node is no longer deflated because
 882  * we're running out of delta_base_cache_limit; we need to re-deflate
 883  * parents, possibly up to the top base.
 884  *
 885  * All deflated objects here are subject to be freed if we exceed
 886  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 887  * just need to make sure the last node is not freed.
 888  */
 889 static void *get_base_data(struct base_data *c)
 890 {
 891         if (!c->data) {
 892                 struct object_entry *obj = c->obj;
 893                 struct base_data **delta = NULL;
 894                 int delta_nr = 0, delta_alloc = 0;
 895
 896                 while (is_delta_type(c->obj->type) && !c->data) {
 897                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 898                         delta[delta_nr++] = c;
 899                         c = c->base;
 900                 }
 901                 if (!delta_nr) {
 902                         c->data = get_data_from_pack(obj);
 903                         c->size = obj->size;
 904                         get_thread_data()->base_cache_used += c->size;
 905                         prune_base_data(c);
 906                 }
 907                 for (; delta_nr > 0; delta_nr--) {
 908                         void *base, *raw;
 909                         c = delta[delta_nr - 1];
 910                         obj = c->obj;
 911                         base = get_base_data(c->base);
 912                         raw = get_data_from_pack(obj);
 913                         c->data = patch_delta(
 914                                 base, c->base->size,
 915                                 raw, obj->size,
 916                                 &c->size);
 917                         free(raw);
 918                         if (!c->data)
 919                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 920                         get_thread_data()->base_cache_used += c->size;
 921                         prune_base_data(c);
 922                 }
 923                 free(delta);
 924         }
 925         return c->data;
 926 }
 927
 928 static void resolve_delta(struct object_entry *delta_obj,
 929                           struct base_data *base, struct base_data *result)
 930 {
 931         void *base_data, *delta_data;
 932
 933         if (show_stat) {
 934                 int i = delta_obj - objects;
 935                 int j = base->obj - objects;
 936                 obj_stat[i].delta_depth = obj_stat[j].delta_depth + 1;
 937                 deepest_delta_lock();
 938                 if (deepest_delta < obj_stat[i].delta_depth)
 939                         deepest_delta = obj_stat[i].delta_depth;
 940                 deepest_delta_unlock();
 941                 obj_stat[i].base_object_no = j;
 942         }
 943         delta_data = get_data_from_pack(delta_obj);
 944         base_data = get_base_data(base);
 945         result->obj = delta_obj;
 946         result->data = patch_delta(base_data, base->size,
 947                                    delta_data, delta_obj->size, &result->size);
 948         free(delta_data);
 949         if (!result->data)
 950                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 951         hash_sha1_file(result->data, result->size,
 952                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 953         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 954                     delta_obj->idx.sha1);
 955         counter_lock();
 956         nr_resolved_deltas++;
 957         counter_unlock();
 958 }
 959
 960 /*
 961  * Standard boolean compare-and-swap: atomically check whether "*type" is
 962  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 963  * and return false.
 964  */
 965 static int compare_and_swap_type(signed char *type,
 966                                  enum object_type want,
 967                                  enum object_type set)
 968 {
 969         enum object_type old;
 970
 971         type_cas_lock();
 972         old = *type;
 973         if (old == want)
 974                 *type = set;
 975         type_cas_unlock();
 976
 977         return old == want;
 978 }
 979
 980 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 981                                                   struct base_data *prev_base)
 982 {
 983         if (base->ref_last == -1 && base->ofs_last == -1) {
 984                 find_ref_delta_children(base->obj->idx.sha1,
 985                                         &base->ref_first, &base->ref_last,
 986                                         OBJ_REF_DELTA);
 987
 988                 find_ofs_delta_children(base->obj->idx.offset,
 989                                         &base->ofs_first, &base->ofs_last,
 990                                         OBJ_OFS_DELTA);
 991
 992                 if (base->ref_last == -1 && base->ofs_last == -1) {
 993                         free(base->data);
 994                         return NULL;
 995                 }
 996
 997                 link_base_data(prev_base, base);
 998         }
 999
1000         if (base->ref_first <= base->ref_last) {
1001                 struct object_entry *child = objects + ref_deltas[base->ref_first].obj_no;
1002                 struct base_data *result = alloc_base_data();
1003
1004                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
1005                                            base->obj->real_type))
1006                         die("BUG: child->real_type != OBJ_REF_DELTA");
1007
1008                 resolve_delta(child, base, result);
1009                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
1010                         free_base_data(base);
1011
1012                 base->ref_first++;
1013                 return result;
1014         }
1015
1016         if (base->ofs_first <= base->ofs_last) {
1017                 struct object_entry *child = objects + ofs_deltas[base->ofs_first].obj_no;
1018                 struct base_data *result = alloc_base_data();
1019
1020                 assert(child->real_type == OBJ_OFS_DELTA);
1021                 child->real_type = base->obj->real_type;
1022                 resolve_delta(child, base, result);
1023                 if (base->ofs_first == base->ofs_last)
1024                         free_base_data(base);
1025
1026                 base->ofs_first++;
1027                 return result;
1028         }
1029
1030         unlink_base_data(base);
1031         return NULL;
1032 }
1033
1034 static void find_unresolved_deltas(struct base_data *base)
1035 {
1036         struct base_data *new_base, *prev_base = NULL;
1037         for (;;) {
1038                 new_base = find_unresolved_deltas_1(base, prev_base);
1039
1040                 if (new_base) {
1041                         prev_base = base;
1042                         base = new_base;
1043                 } else {
1044                         free(base);
1045                         base = prev_base;
1046                         if (!base)
1047                                 return;
1048                         prev_base = base->base;
1049                 }
1050         }
1051 }
1052
1053 static int compare_ofs_delta_entry(const void *a, const void *b)
1054 {
1055         const struct ofs_delta_entry *delta_a = a;
1056         const struct ofs_delta_entry *delta_b = b;
1057
1058         return delta_a->offset < delta_b->offset ? -1 :
1059                delta_a->offset > delta_b->offset ?  1 :
1060                0;
1061 }
1062
1063 static int compare_ref_delta_entry(const void *a, const void *b)
1064 {
1065         const struct ref_delta_entry *delta_a = a;
1066         const struct ref_delta_entry *delta_b = b;
1067
1068         return hashcmp(delta_a->sha1, delta_b->sha1);
1069 }
1070
1071 static void resolve_base(struct object_entry *obj)
1072 {
1073         struct base_data *base_obj = alloc_base_data();
1074         base_obj->obj = obj;
1075         base_obj->data = NULL;
1076         find_unresolved_deltas(base_obj);
1077 }
1078
1079 #ifndef NO_PTHREADS
1080 static void *threaded_second_pass(void *data)
1081 {
1082         set_thread_data(data);
1083         for (;;) {
1084                 int i;
1085                 counter_lock();
1086                 display_progress(progress, nr_resolved_deltas);
1087                 counter_unlock();
1088                 work_lock();
1089                 while (nr_dispatched < nr_objects &&
1090                        is_delta_type(objects[nr_dispatched].type))
1091                         nr_dispatched++;
1092                 if (nr_dispatched >= nr_objects) {
1093                         work_unlock();
1094                         break;
1095                 }
1096                 i = nr_dispatched++;
1097                 work_unlock();
1098
1099                 resolve_base(&objects[i]);
1100         }
1101         return NULL;
1102 }
1103 #endif
1104
1105 /*
1106  * First pass:
1107  * - find locations of all objects;
1108  * - calculate SHA1 of all non-delta objects;
1109  * - remember base (SHA1 or offset) for all deltas.
1110  */
1111 static void parse_pack_objects(unsigned char *sha1)
1112 {
1113         int i, nr_delays = 0;
1114         struct ofs_delta_entry *ofs_delta = ofs_deltas;
1115         unsigned char ref_delta_sha1[20];
1116         struct stat st;
1117
1118         if (verbose)
1119                 progress = start_progress(
1120                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1121                                 nr_objects);
1122         for (i = 0; i < nr_objects; i++) {
1123                 struct object_entry *obj = &objects[i];
1124                 void *data = unpack_raw_entry(obj, &ofs_delta->offset,
1125                                               ref_delta_sha1, obj->idx.sha1);
1126                 obj->real_type = obj->type;
1127                 if (obj->type == OBJ_OFS_DELTA) {
1128                         nr_ofs_deltas++;
1129                         ofs_delta->obj_no = i;
1130                         ofs_delta++;
1131                 } else if (obj->type == OBJ_REF_DELTA) {
1132                         ALLOC_GROW(ref_deltas, nr_ref_deltas + 1, ref_deltas_alloc);
1133                         hashcpy(ref_deltas[nr_ref_deltas].sha1, ref_delta_sha1);
1134                         ref_deltas[nr_ref_deltas].obj_no = i;
1135                         nr_ref_deltas++;
1136                 } else if (!data) {
1137                         /* large blobs, check later */
1138                         obj->real_type = OBJ_BAD;
1139                         nr_delays++;
1140                 } else
1141                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1142                 free(data);
1143                 display_progress(progress, i+1);
1144         }
1145         objects[i].idx.offset = consumed_bytes;
1146         stop_progress(&progress);
1147
1148         /* Check pack integrity */
1149         flush();
1150         git_SHA1_Final(sha1, &input_ctx);
1151         if (hashcmp(fill(20), sha1))
1152                 die(_("pack is corrupted (SHA1 mismatch)"));
1153         use(20);
1154
1155         /* If input_fd is a file, we should have reached its end now. */
1156         if (fstat(input_fd, &st))
1157                 die_errno(_("cannot fstat packfile"));
1158         if (S_ISREG(st.st_mode) &&
1159                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1160                 die(_("pack has junk at the end"));
1161
1162         for (i = 0; i < nr_objects; i++) {
1163                 struct object_entry *obj = &objects[i];
1164                 if (obj->real_type != OBJ_BAD)
1165                         continue;
1166                 obj->real_type = obj->type;
1167                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1168                 nr_delays--;
1169         }
1170         if (nr_delays)
1171                 die(_("confusion beyond insanity in parse_pack_objects()"));
1172 }
1173
1174 /*
1175  * Second pass:
1176  * - for all non-delta objects, look if it is used as a base for
1177  *   deltas;
1178  * - if used as a base, uncompress the object and apply all deltas,
1179  *   recursively checking if the resulting object is used as a base
1180  *   for some more deltas.
1181  */
1182 static void resolve_deltas(void)
1183 {
1184         int i;
1185
1186         if (!nr_ofs_deltas && !nr_ref_deltas)
1187                 return;
1188
1189         /* Sort deltas by base SHA1/offset for fast searching */
1190         qsort(ofs_deltas, nr_ofs_deltas, sizeof(struct ofs_delta_entry),
1191               compare_ofs_delta_entry);
1192         qsort(ref_deltas, nr_ref_deltas, sizeof(struct ref_delta_entry),
1193               compare_ref_delta_entry);
1194
1195         if (verbose || show_resolving_progress)
1196                 progress = start_progress(_("Resolving deltas"),
1197                                           nr_ref_deltas + nr_ofs_deltas);
1198
1199 #ifndef NO_PTHREADS
1200         nr_dispatched = 0;
1201         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1202                 init_thread();
1203                 for (i = 0; i < nr_threads; i++) {
1204                         int ret = pthread_create(&thread_data[i].thread, NULL,
1205                                                  threaded_second_pass, thread_data + i);
1206                         if (ret)
1207                                 die(_("unable to create thread: %s"),
1208                                     strerror(ret));
1209                 }
1210                 for (i = 0; i < nr_threads; i++)
1211                         pthread_join(thread_data[i].thread, NULL);
1212                 cleanup_thread();
1213                 return;
1214         }
1215 #endif
1216
1217         for (i = 0; i < nr_objects; i++) {
1218                 struct object_entry *obj = &objects[i];
1219
1220                 if (is_delta_type(obj->type))
1221                         continue;
1222                 resolve_base(obj);
1223                 display_progress(progress, nr_resolved_deltas);
1224         }
1225 }
1226
1227 /*
1228  * Third pass:
1229  * - append objects to convert thin pack to full pack if required
1230  * - write the final 20-byte SHA-1
1231  */
1232 static void fix_unresolved_deltas(struct sha1file *f);
1233 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1234 {
1235         if (nr_ref_deltas + nr_ofs_deltas == nr_resolved_deltas) {
1236                 stop_progress(&progress);
1237                 /* Flush remaining pack final 20-byte SHA1. */
1238                 flush();
1239                 return;
1240         }
1241
1242         if (fix_thin_pack) {
1243                 struct sha1file *f;
1244                 unsigned char read_sha1[20], tail_sha1[20];
1245                 struct strbuf msg = STRBUF_INIT;
1246                 int nr_unresolved = nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas;
1247                 int nr_objects_initial = nr_objects;
1248                 if (nr_unresolved <= 0)
1249                         die(_("confusion beyond insanity"));
1250                 REALLOC_ARRAY(objects, nr_objects + nr_unresolved + 1);
1251                 memset(objects + nr_objects + 1, 0,
1252                        nr_unresolved * sizeof(*objects));
1253                 f = sha1fd(output_fd, curr_pack);
1254                 fix_unresolved_deltas(f);
1255                 strbuf_addf(&msg, Q_("completed with %d local object",
1256                                      "completed with %d local objects",
1257                                      nr_objects - nr_objects_initial),
1258                             nr_objects - nr_objects_initial);
1259                 stop_progress_msg(&progress, msg.buf);
1260                 strbuf_release(&msg);
1261                 sha1close(f, tail_sha1, 0);
1262                 hashcpy(read_sha1, pack_sha1);
1263                 fixup_pack_header_footer(output_fd, pack_sha1,
1264                                          curr_pack, nr_objects,
1265                                          read_sha1, consumed_bytes-20);
1266                 if (hashcmp(read_sha1, tail_sha1) != 0)
1267                         die(_("Unexpected tail checksum for %s "
1268                               "(disk corruption?)"), curr_pack);
1269         }
1270         if (nr_ofs_deltas + nr_ref_deltas != nr_resolved_deltas)
1271                 die(Q_("pack has %d unresolved delta",
1272                        "pack has %d unresolved deltas",
1273                        nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas),
1274                     nr_ofs_deltas + nr_ref_deltas - nr_resolved_deltas);
1275 }
1276
1277 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1278 {
1279         git_zstream stream;
1280         int status;
1281         unsigned char outbuf[4096];
1282
1283         git_deflate_init(&stream, zlib_compression_level);
1284         stream.next_in = in;
1285         stream.avail_in = size;
1286
1287         do {
1288                 stream.next_out = outbuf;
1289                 stream.avail_out = sizeof(outbuf);
1290                 status = git_deflate(&stream, Z_FINISH);
1291                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1292         } while (status == Z_OK);
1293
1294         if (status != Z_STREAM_END)
1295                 die(_("unable to deflate appended object (%d)"), status);
1296         size = stream.total_out;
1297         git_deflate_end(&stream);
1298         return size;
1299 }
1300
1301 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1302                                const unsigned char *sha1, void *buf,
1303                                unsigned long size, enum object_type type)
1304 {
1305         struct object_entry *obj = &objects[nr_objects++];
1306         unsigned char header[10];
1307         unsigned long s = size;
1308         int n = 0;
1309         unsigned char c = (type << 4) | (s & 15);
1310         s >>= 4;
1311         while (s) {
1312                 header[n++] = c | 0x80;
1313                 c = s & 0x7f;
1314                 s >>= 7;
1315         }
1316         header[n++] = c;
1317         crc32_begin(f);
1318         sha1write(f, header, n);
1319         obj[0].size = size;
1320         obj[0].hdr_size = n;
1321         obj[0].type = type;
1322         obj[0].real_type = type;
1323         obj[1].idx.offset = obj[0].idx.offset + n;
1324         obj[1].idx.offset += write_compressed(f, buf, size);
1325         obj[0].idx.crc32 = crc32_end(f);
1326         sha1flush(f);
1327         hashcpy(obj->idx.sha1, sha1);
1328         return obj;
1329 }
1330
1331 static int delta_pos_compare(const void *_a, const void *_b)
1332 {
1333         struct ref_delta_entry *a = *(struct ref_delta_entry **)_a;
1334         struct ref_delta_entry *b = *(struct ref_delta_entry **)_b;
1335         return a->obj_no - b->obj_no;
1336 }
1337
1338 static void fix_unresolved_deltas(struct sha1file *f)
1339 {
1340         struct ref_delta_entry **sorted_by_pos;
1341         int i;
1342
1343         /*
1344          * Since many unresolved deltas may well be themselves base objects
1345          * for more unresolved deltas, we really want to include the
1346          * smallest number of base objects that would cover as much delta
1347          * as possible by picking the
1348          * trunc deltas first, allowing for other deltas to resolve without
1349          * additional base objects.  Since most base objects are to be found
1350          * before deltas depending on them, a good heuristic is to start
1351          * resolving deltas in the same order as their position in the pack.
1352          */
1353         ALLOC_ARRAY(sorted_by_pos, nr_ref_deltas);
1354         for (i = 0; i < nr_ref_deltas; i++)
1355                 sorted_by_pos[i] = &ref_deltas[i];
1356         qsort(sorted_by_pos, nr_ref_deltas, sizeof(*sorted_by_pos), delta_pos_compare);
1357
1358         for (i = 0; i < nr_ref_deltas; i++) {
1359                 struct ref_delta_entry *d = sorted_by_pos[i];
1360                 enum object_type type;
1361                 struct base_data *base_obj = alloc_base_data();
1362
1363                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1364                         continue;
1365                 base_obj->data = read_sha1_file(d->sha1, &type, &base_obj->size);
1366                 if (!base_obj->data)
1367                         continue;
1368
1369                 if (check_sha1_signature(d->sha1, base_obj->data,
1370                                 base_obj->size, typename(type)))
1371                         die(_("local object %s is corrupt"), sha1_to_hex(d->sha1));
1372                 base_obj->obj = append_obj_to_pack(f, d->sha1,
1373                                         base_obj->data, base_obj->size, type);
1374                 find_unresolved_deltas(base_obj);
1375                 display_progress(progress, nr_resolved_deltas);
1376         }
1377         free(sorted_by_pos);
1378 }
1379
1380 static void final(const char *final_pack_name, const char *curr_pack_name,
1381                   const char *final_index_name, const char *curr_index_name,
1382                   const char *keep_name, const char *keep_msg,
1383                   unsigned char *sha1)
1384 {
1385         const char *report = "pack";
1386         char name[PATH_MAX];
1387         int err;
1388
1389         if (!from_stdin) {
1390                 close(input_fd);
1391         } else {
1392                 fsync_or_die(output_fd, curr_pack_name);
1393                 err = close(output_fd);
1394                 if (err)
1395                         die_errno(_("error while closing pack file"));
1396         }
1397
1398         if (keep_msg) {
1399                 int keep_fd, keep_msg_len = strlen(keep_msg);
1400
1401                 if (!keep_name)
1402                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1403                 else
1404                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1405
1406                 if (keep_fd < 0) {
1407                         if (errno != EEXIST)
1408                                 die_errno(_("cannot write keep file '%s'"),
1409                                           keep_name ? keep_name : name);
1410                 } else {
1411                         if (keep_msg_len > 0) {
1412                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1413                                 write_or_die(keep_fd, "\n", 1);
1414                         }
1415                         if (close(keep_fd) != 0)
1416                                 die_errno(_("cannot close written keep file '%s'"),
1417                                           keep_name ? keep_name : name);
1418                         report = "keep";
1419                 }
1420         }
1421
1422         if (final_pack_name != curr_pack_name) {
1423                 if (!final_pack_name) {
1424                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1425                                  get_object_directory(), sha1_to_hex(sha1));
1426                         final_pack_name = name;
1427                 }
1428                 if (finalize_object_file(curr_pack_name, final_pack_name))
1429                         die(_("cannot store pack file"));
1430         } else if (from_stdin)
1431                 chmod(final_pack_name, 0444);
1432
1433         if (final_index_name != curr_index_name) {
1434                 if (!final_index_name) {
1435                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1436                                  get_object_directory(), sha1_to_hex(sha1));
1437                         final_index_name = name;
1438                 }
1439                 if (finalize_object_file(curr_index_name, final_index_name))
1440                         die(_("cannot store index file"));
1441         } else
1442                 chmod(final_index_name, 0444);
1443
1444         if (!from_stdin) {
1445                 printf("%s\n", sha1_to_hex(sha1));
1446         } else {
1447                 char buf[48];
1448                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1449                                    report, sha1_to_hex(sha1));
1450                 write_or_die(1, buf, len);
1451
1452                 /*
1453                  * Let's just mimic git-unpack-objects here and write
1454                  * the last part of the input buffer to stdout.
1455                  */
1456                 while (input_len) {
1457                         err = xwrite(1, input_buffer + input_offset, input_len);
1458                         if (err <= 0)
1459                                 break;
1460                         input_len -= err;
1461                         input_offset += err;
1462                 }
1463         }
1464 }
1465
1466 static int git_index_pack_config(const char *k, const char *v, void *cb)
1467 {
1468         struct pack_idx_option *opts = cb;
1469
1470         if (!strcmp(k, "pack.indexversion")) {
1471                 opts->version = git_config_int(k, v);
1472                 if (opts->version > 2)
1473                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1474                 return 0;
1475         }
1476         if (!strcmp(k, "pack.threads")) {
1477                 nr_threads = git_config_int(k, v);
1478                 if (nr_threads < 0)
1479                         die(_("invalid number of threads specified (%d)"),
1480                             nr_threads);
1481 #ifdef NO_PTHREADS
1482                 if (nr_threads != 1)
1483                         warning(_("no threads support, ignoring %s"), k);
1484                 nr_threads = 1;
1485 #endif
1486                 return 0;
1487         }
1488         return git_default_config(k, v, cb);
1489 }
1490
1491 static int cmp_uint32(const void *a_, const void *b_)
1492 {
1493         uint32_t a = *((uint32_t *)a_);
1494         uint32_t b = *((uint32_t *)b_);
1495
1496         return (a < b) ? -1 : (a != b);
1497 }
1498
1499 static void read_v2_anomalous_offsets(struct packed_git *p,
1500                                       struct pack_idx_option *opts)
1501 {
1502         const uint32_t *idx1, *idx2;
1503         uint32_t i;
1504
1505         /* The address of the 4-byte offset table */
1506         idx1 = (((const uint32_t *)p->index_data)
1507                 + 2 /* 8-byte header */
1508                 + 256 /* fan out */
1509                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1510                 + p->num_objects /* CRC32 table */
1511                 );
1512
1513         /* The address of the 8-byte offset table */
1514         idx2 = idx1 + p->num_objects;
1515
1516         for (i = 0; i < p->num_objects; i++) {
1517                 uint32_t off = ntohl(idx1[i]);
1518                 if (!(off & 0x80000000))
1519                         continue;
1520                 off = off & 0x7fffffff;
1521                 check_pack_index_ptr(p, &idx2[off * 2]);
1522                 if (idx2[off * 2])
1523                         continue;
1524                 /*
1525                  * The real offset is ntohl(idx2[off * 2]) in high 4
1526                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1527                  * octets.  But idx2[off * 2] is Zero!!!
1528                  */
1529                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1530                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1531         }
1532
1533         if (1 < opts->anomaly_nr)
1534                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), 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
1718                                 usage(index_pack_usage);
1719                         continue;
1720                 }
1721
1722                 if (pack_name)
1723                         usage(index_pack_usage);
1724                 pack_name = arg;
1725         }
1726
1727         if (!pack_name && !from_stdin)
1728                 usage(index_pack_usage);
1729         if (fix_thin_pack && !from_stdin)
1730                 die(_("--fix-thin cannot be used without --stdin"));
1731         if (!index_name && pack_name)
1732                 index_name = derive_filename(pack_name, ".idx", &index_name_buf);
1733         if (keep_msg && !keep_name && pack_name)
1734                 keep_name = derive_filename(pack_name, ".keep", &keep_name_buf);
1735
1736         if (verify) {
1737                 if (!index_name)
1738                         die(_("--verify with no packfile name given"));
1739                 read_idx_option(&opts, index_name);
1740                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1741         }
1742         if (strict)
1743                 opts.flags |= WRITE_IDX_STRICT;
1744
1745 #ifndef NO_PTHREADS
1746         if (!nr_threads) {
1747                 nr_threads = online_cpus();
1748                 /* An experiment showed that more threads does not mean faster */
1749                 if (nr_threads > 3)
1750                         nr_threads = 3;
1751         }
1752 #endif
1753
1754         curr_pack = open_pack_file(pack_name);
1755         parse_pack_header();
1756         objects = xcalloc(st_add(nr_objects, 1), sizeof(struct object_entry));
1757         if (show_stat)
1758                 obj_stat = xcalloc(st_add(nr_objects, 1), sizeof(struct object_stat));
1759         ofs_deltas = xcalloc(nr_objects, sizeof(struct ofs_delta_entry));
1760         parse_pack_objects(pack_sha1);
1761         if (report_end_of_input)
1762                 write_in_full(2, "\0", 1);
1763         resolve_deltas();
1764         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1765         free(ofs_deltas);
1766         free(ref_deltas);
1767         if (strict)
1768                 foreign_nr = check_objects();
1769
1770         if (show_stat)
1771                 show_pack_info(stat_only);
1772
1773         ALLOC_ARRAY(idx_objects, nr_objects);
1774         for (i = 0; i < nr_objects; i++)
1775                 idx_objects[i] = &objects[i].idx;
1776         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1777         free(idx_objects);
1778
1779         if (!verify)
1780                 final(pack_name, curr_pack,
1781                       index_name, curr_index,
1782                       keep_name, keep_msg,
1783                       pack_sha1);
1784         else
1785                 close(input_fd);
1786         free(objects);
1787         strbuf_release(&index_name_buf);
1788         strbuf_release(&keep_name_buf);
1789         if (pack_name == NULL)
1790                 free((void *) curr_pack);
1791         if (index_name == NULL)
1792                 free((void *) curr_index);
1793
1794         /*
1795          * Let the caller know this pack is not self contained
1796          */
1797         if (check_self_contained_and_connected && foreign_nr)
1798                 return 1;
1799
1800         return 0;
1801 }