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 int hdr_size;
  22         enum object_type type;
  23         enum object_type real_type;
  24         unsigned delta_depth;
  25         int base_object_no;
  26 };
  27
  28 union delta_base {
  29         unsigned char sha1[20];
  30         off_t offset;
  31 };
  32
  33 struct base_data {
  34         struct base_data *base;
  35         struct base_data *child;
  36         struct object_entry *obj;
  37         void *data;
  38         unsigned long size;
  39         int ref_first, ref_last;
  40         int ofs_first, ofs_last;
  41 };
  42
  43 #if !defined(NO_PTHREADS) && defined(NO_THREAD_SAFE_PREAD)
  44 /* pread() emulation is not thread-safe. Disable threading. */
  45 #define NO_PTHREADS
  46 #endif
  47
  48 struct thread_local {
  49 #ifndef NO_PTHREADS
  50         pthread_t thread;
  51 #endif
  52         struct base_data *base_cache;
  53         size_t base_cache_used;
  54 };
  55
  56 /*
  57  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  58  * to memcmp() only the first 20 bytes.
  59  */
  60 #define UNION_BASE_SZ   20
  61
  62 #define FLAG_LINK (1u<<20)
  63 #define FLAG_CHECKED (1u<<21)
  64
  65 struct delta_entry {
  66         union delta_base base;
  67         int obj_no;
  68 };
  69
  70 static struct object_entry *objects;
  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, pack_fd;
  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         init_recursive_mutex(&read_mutex);
 142         pthread_mutex_init(&counter_mutex, NULL);
 143         pthread_mutex_init(&work_mutex, NULL);
 144         pthread_mutex_init(&type_cas_mutex, NULL);
 145         if (show_stat)
 146                 pthread_mutex_init(&deepest_delta_mutex, NULL);
 147         pthread_key_create(&key, NULL);
 148         thread_data = xcalloc(nr_threads, sizeof(*thread_data));
 149         threads_active = 1;
 150 }
 151
 152 static void cleanup_thread(void)
 153 {
 154         if (!threads_active)
 155                 return;
 156         threads_active = 0;
 157         pthread_mutex_destroy(&read_mutex);
 158         pthread_mutex_destroy(&counter_mutex);
 159         pthread_mutex_destroy(&work_mutex);
 160         pthread_mutex_destroy(&type_cas_mutex);
 161         if (show_stat)
 162                 pthread_mutex_destroy(&deepest_delta_mutex);
 163         pthread_key_delete(key);
 164         free(thread_data);
 165 }
 166
 167 #else
 168
 169 #define read_lock()
 170 #define read_unlock()
 171
 172 #define counter_lock()
 173 #define counter_unlock()
 174
 175 #define work_lock()
 176 #define work_unlock()
 177
 178 #define deepest_delta_lock()
 179 #define deepest_delta_unlock()
 180
 181 #endif
 182
 183
 184 static int mark_link(struct object *obj, int type, void *data)
 185 {
 186         if (!obj)
 187                 return -1;
 188
 189         if (type != OBJ_ANY && obj->type != type)
 190                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
 191
 192         obj->flags |= FLAG_LINK;
 193         return 0;
 194 }
 195
 196 /* The content of each linked object must have been checked
 197    or it must be already present in the object database */
 198 static unsigned check_object(struct object *obj)
 199 {
 200         if (!obj)
 201                 return 0;
 202
 203         if (!(obj->flags & FLAG_LINK))
 204                 return 0;
 205
 206         if (!(obj->flags & FLAG_CHECKED)) {
 207                 unsigned long size;
 208                 int type = sha1_object_info(obj->sha1, &size);
 209                 if (type != obj->type || type <= 0)
 210                         die(_("object of unexpected type"));
 211                 obj->flags |= FLAG_CHECKED;
 212                 return 1;
 213         }
 214
 215         return 0;
 216 }
 217
 218 static unsigned check_objects(void)
 219 {
 220         unsigned i, max, foreign_nr = 0;
 221
 222         max = get_max_object_index();
 223         for (i = 0; i < max; i++)
 224                 foreign_nr += check_object(get_indexed_object(i));
 225         return foreign_nr;
 226 }
 227
 228
 229 /* Discard current buffer used content. */
 230 static void flush(void)
 231 {
 232         if (input_offset) {
 233                 if (output_fd >= 0)
 234                         write_or_die(output_fd, input_buffer, input_offset);
 235                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 236                 memmove(input_buffer, input_buffer + input_offset, input_len);
 237                 input_offset = 0;
 238         }
 239 }
 240
 241 /*
 242  * Make sure at least "min" bytes are available in the buffer, and
 243  * return the pointer to the buffer.
 244  */
 245 static void *fill(int min)
 246 {
 247         if (min <= input_len)
 248                 return input_buffer + input_offset;
 249         if (min > sizeof(input_buffer))
 250                 die(Q_("cannot fill %d byte",
 251                        "cannot fill %d bytes",
 252                        min),
 253                     min);
 254         flush();
 255         do {
 256                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 257                                 sizeof(input_buffer) - input_len);
 258                 if (ret <= 0) {
 259                         if (!ret)
 260                                 die(_("early EOF"));
 261                         die_errno(_("read error on input"));
 262                 }
 263                 input_len += ret;
 264                 if (from_stdin)
 265                         display_throughput(progress, consumed_bytes + input_len);
 266         } while (input_len < min);
 267         return input_buffer;
 268 }
 269
 270 static void use(int bytes)
 271 {
 272         if (bytes > input_len)
 273                 die(_("used more bytes than were available"));
 274         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 275         input_len -= bytes;
 276         input_offset += bytes;
 277
 278         /* make sure off_t is sufficiently large not to wrap */
 279         if (signed_add_overflows(consumed_bytes, bytes))
 280                 die(_("pack too large for current definition of off_t"));
 281         consumed_bytes += bytes;
 282 }
 283
 284 static const char *open_pack_file(const char *pack_name)
 285 {
 286         if (from_stdin) {
 287                 input_fd = 0;
 288                 if (!pack_name) {
 289                         static char tmp_file[PATH_MAX];
 290                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 291                                                 "pack/tmp_pack_XXXXXX");
 292                         pack_name = xstrdup(tmp_file);
 293                 } else
 294                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 295                 if (output_fd < 0)
 296                         die_errno(_("unable to create '%s'"), pack_name);
 297                 pack_fd = output_fd;
 298         } else {
 299                 input_fd = open(pack_name, O_RDONLY);
 300                 if (input_fd < 0)
 301                         die_errno(_("cannot open packfile '%s'"), pack_name);
 302                 output_fd = -1;
 303                 pack_fd = input_fd;
 304         }
 305         git_SHA1_Init(&input_ctx);
 306         return pack_name;
 307 }
 308
 309 static void parse_pack_header(void)
 310 {
 311         struct pack_header *hdr = fill(sizeof(struct pack_header));
 312
 313         /* Header consistency check */
 314         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 315                 die(_("pack signature mismatch"));
 316         if (!pack_version_ok(hdr->hdr_version))
 317                 die(_("pack version %"PRIu32" unsupported"),
 318                         ntohl(hdr->hdr_version));
 319
 320         nr_objects = ntohl(hdr->hdr_entries);
 321         use(sizeof(struct pack_header));
 322 }
 323
 324 static NORETURN void bad_object(unsigned long offset, const char *format,
 325                        ...) __attribute__((format (printf, 2, 3)));
 326
 327 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 328 {
 329         va_list params;
 330         char buf[1024];
 331
 332         va_start(params, format);
 333         vsnprintf(buf, sizeof(buf), format, params);
 334         va_end(params);
 335         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 336 }
 337
 338 static inline struct thread_local *get_thread_data(void)
 339 {
 340 #ifndef NO_PTHREADS
 341         if (threads_active)
 342                 return pthread_getspecific(key);
 343         assert(!threads_active &&
 344                "This should only be reached when all threads are gone");
 345 #endif
 346         return &nothread_data;
 347 }
 348
 349 #ifndef NO_PTHREADS
 350 static void set_thread_data(struct thread_local *data)
 351 {
 352         if (threads_active)
 353                 pthread_setspecific(key, data);
 354 }
 355 #endif
 356
 357 static struct base_data *alloc_base_data(void)
 358 {
 359         struct base_data *base = xmalloc(sizeof(struct base_data));
 360         memset(base, 0, sizeof(*base));
 361         base->ref_last = -1;
 362         base->ofs_last = -1;
 363         return base;
 364 }
 365
 366 static void free_base_data(struct base_data *c)
 367 {
 368         if (c->data) {
 369                 free(c->data);
 370                 c->data = NULL;
 371                 get_thread_data()->base_cache_used -= c->size;
 372         }
 373 }
 374
 375 static void prune_base_data(struct base_data *retain)
 376 {
 377         struct base_data *b;
 378         struct thread_local *data = get_thread_data();
 379         for (b = data->base_cache;
 380              data->base_cache_used > delta_base_cache_limit && b;
 381              b = b->child) {
 382                 if (b->data && b != retain)
 383                         free_base_data(b);
 384         }
 385 }
 386
 387 static void link_base_data(struct base_data *base, struct base_data *c)
 388 {
 389         if (base)
 390                 base->child = c;
 391         else
 392                 get_thread_data()->base_cache = c;
 393
 394         c->base = base;
 395         c->child = NULL;
 396         if (c->data)
 397                 get_thread_data()->base_cache_used += c->size;
 398         prune_base_data(c);
 399 }
 400
 401 static void unlink_base_data(struct base_data *c)
 402 {
 403         struct base_data *base = c->base;
 404         if (base)
 405                 base->child = NULL;
 406         else
 407                 get_thread_data()->base_cache = NULL;
 408         free_base_data(c);
 409 }
 410
 411 static int is_delta_type(enum object_type type)
 412 {
 413         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 414 }
 415
 416 static void *unpack_entry_data(unsigned long offset, unsigned long size,
 417                                enum object_type type, unsigned char *sha1)
 418 {
 419         static char fixed_buf[8192];
 420         int status;
 421         git_zstream stream;
 422         void *buf;
 423         git_SHA_CTX c;
 424         char hdr[32];
 425         int hdrlen;
 426
 427         if (!is_delta_type(type)) {
 428                 hdrlen = sprintf(hdr, "%s %lu", typename(type), size) + 1;
 429                 git_SHA1_Init(&c);
 430                 git_SHA1_Update(&c, hdr, hdrlen);
 431         } else
 432                 sha1 = NULL;
 433         if (type == OBJ_BLOB && size > big_file_threshold)
 434                 buf = fixed_buf;
 435         else
 436                 buf = xmalloc(size);
 437
 438         memset(&stream, 0, sizeof(stream));
 439         git_inflate_init(&stream);
 440         stream.next_out = buf;
 441         stream.avail_out = buf == fixed_buf ? sizeof(fixed_buf) : size;
 442
 443         do {
 444                 unsigned char *last_out = stream.next_out;
 445                 stream.next_in = fill(1);
 446                 stream.avail_in = input_len;
 447                 status = git_inflate(&stream, 0);
 448                 use(input_len - stream.avail_in);
 449                 if (sha1)
 450                         git_SHA1_Update(&c, last_out, stream.next_out - last_out);
 451                 if (buf == fixed_buf) {
 452                         stream.next_out = buf;
 453                         stream.avail_out = sizeof(fixed_buf);
 454                 }
 455         } while (status == Z_OK);
 456         if (stream.total_out != size || status != Z_STREAM_END)
 457                 bad_object(offset, _("inflate returned %d"), status);
 458         git_inflate_end(&stream);
 459         if (sha1)
 460                 git_SHA1_Final(sha1, &c);
 461         return buf == fixed_buf ? NULL : buf;
 462 }
 463
 464 static void *unpack_raw_entry(struct object_entry *obj,
 465                               union delta_base *delta_base,
 466                               unsigned char *sha1)
 467 {
 468         unsigned char *p;
 469         unsigned long size, c;
 470         off_t base_offset;
 471         unsigned shift;
 472         void *data;
 473
 474         obj->idx.offset = consumed_bytes;
 475         input_crc32 = crc32(0, NULL, 0);
 476
 477         p = fill(1);
 478         c = *p;
 479         use(1);
 480         obj->type = (c >> 4) & 7;
 481         size = (c & 15);
 482         shift = 4;
 483         while (c & 0x80) {
 484                 p = fill(1);
 485                 c = *p;
 486                 use(1);
 487                 size += (c & 0x7f) << shift;
 488                 shift += 7;
 489         }
 490         obj->size = size;
 491
 492         switch (obj->type) {
 493         case OBJ_REF_DELTA:
 494                 hashcpy(delta_base->sha1, fill(20));
 495                 use(20);
 496                 break;
 497         case OBJ_OFS_DELTA:
 498                 memset(delta_base, 0, sizeof(*delta_base));
 499                 p = fill(1);
 500                 c = *p;
 501                 use(1);
 502                 base_offset = c & 127;
 503                 while (c & 128) {
 504                         base_offset += 1;
 505                         if (!base_offset || MSB(base_offset, 7))
 506                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 507                         p = fill(1);
 508                         c = *p;
 509                         use(1);
 510                         base_offset = (base_offset << 7) + (c & 127);
 511                 }
 512                 delta_base->offset = obj->idx.offset - base_offset;
 513                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 514                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 515                 break;
 516         case OBJ_COMMIT:
 517         case OBJ_TREE:
 518         case OBJ_BLOB:
 519         case OBJ_TAG:
 520                 break;
 521         default:
 522                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 523         }
 524         obj->hdr_size = consumed_bytes - obj->idx.offset;
 525
 526         data = unpack_entry_data(obj->idx.offset, obj->size, obj->type, sha1);
 527         obj->idx.crc32 = input_crc32;
 528         return data;
 529 }
 530
 531 static void *unpack_data(struct object_entry *obj,
 532                          int (*consume)(const unsigned char *, unsigned long, void *),
 533                          void *cb_data)
 534 {
 535         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 536         unsigned long len = obj[1].idx.offset - from;
 537         unsigned char *data, *inbuf;
 538         git_zstream stream;
 539         int status;
 540
 541         data = xmalloc(consume ? 64*1024 : obj->size);
 542         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 543
 544         memset(&stream, 0, sizeof(stream));
 545         git_inflate_init(&stream);
 546         stream.next_out = data;
 547         stream.avail_out = consume ? 64*1024 : obj->size;
 548
 549         do {
 550                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 551                 n = pread(pack_fd, inbuf, n, from);
 552                 if (n < 0)
 553                         die_errno(_("cannot pread pack file"));
 554                 if (!n)
 555                         die(Q_("premature end of pack file, %lu byte missing",
 556                                "premature end of pack file, %lu bytes missing",
 557                                len),
 558                             len);
 559                 from += n;
 560                 len -= n;
 561                 stream.next_in = inbuf;
 562                 stream.avail_in = n;
 563                 if (!consume)
 564                         status = git_inflate(&stream, 0);
 565                 else {
 566                         do {
 567                                 status = git_inflate(&stream, 0);
 568                                 if (consume(data, stream.next_out - data, cb_data)) {
 569                                         free(inbuf);
 570                                         free(data);
 571                                         return NULL;
 572                                 }
 573                                 stream.next_out = data;
 574                                 stream.avail_out = 64*1024;
 575                         } while (status == Z_OK && stream.avail_in);
 576                 }
 577         } while (len && status == Z_OK && !stream.avail_in);
 578
 579         /* This has been inflated OK when first encountered, so... */
 580         if (status != Z_STREAM_END || stream.total_out != obj->size)
 581                 die(_("serious inflate inconsistency"));
 582
 583         git_inflate_end(&stream);
 584         free(inbuf);
 585         if (consume) {
 586                 free(data);
 587                 data = NULL;
 588         }
 589         return data;
 590 }
 591
 592 static void *get_data_from_pack(struct object_entry *obj)
 593 {
 594         return unpack_data(obj, NULL, NULL);
 595 }
 596
 597 static int compare_delta_bases(const union delta_base *base1,
 598                                const union delta_base *base2,
 599                                enum object_type type1,
 600                                enum object_type type2)
 601 {
 602         int cmp = type1 - type2;
 603         if (cmp)
 604                 return cmp;
 605         return memcmp(base1, base2, UNION_BASE_SZ);
 606 }
 607
 608 static int find_delta(const union delta_base *base, enum object_type type)
 609 {
 610         int first = 0, last = nr_deltas;
 611
 612         while (first < last) {
 613                 int next = (first + last) / 2;
 614                 struct delta_entry *delta = &deltas[next];
 615                 int cmp;
 616
 617                 cmp = compare_delta_bases(base, &delta->base,
 618                                           type, objects[delta->obj_no].type);
 619                 if (!cmp)
 620                         return next;
 621                 if (cmp < 0) {
 622                         last = next;
 623                         continue;
 624                 }
 625                 first = next+1;
 626         }
 627         return -first-1;
 628 }
 629
 630 static void find_delta_children(const union delta_base *base,
 631                                 int *first_index, int *last_index,
 632                                 enum object_type type)
 633 {
 634         int first = find_delta(base, type);
 635         int last = first;
 636         int end = nr_deltas - 1;
 637
 638         if (first < 0) {
 639                 *first_index = 0;
 640                 *last_index = -1;
 641                 return;
 642         }
 643         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 644                 --first;
 645         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 646                 ++last;
 647         *first_index = first;
 648         *last_index = last;
 649 }
 650
 651 struct compare_data {
 652         struct object_entry *entry;
 653         struct git_istream *st;
 654         unsigned char *buf;
 655         unsigned long buf_size;
 656 };
 657
 658 static int compare_objects(const unsigned char *buf, unsigned long size,
 659                            void *cb_data)
 660 {
 661         struct compare_data *data = cb_data;
 662
 663         if (data->buf_size < size) {
 664                 free(data->buf);
 665                 data->buf = xmalloc(size);
 666                 data->buf_size = size;
 667         }
 668
 669         while (size) {
 670                 ssize_t len = read_istream(data->st, data->buf, size);
 671                 if (len == 0)
 672                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 673                             sha1_to_hex(data->entry->idx.sha1));
 674                 if (len < 0)
 675                         die(_("unable to read %s"),
 676                             sha1_to_hex(data->entry->idx.sha1));
 677                 if (memcmp(buf, data->buf, len))
 678                         die(_("SHA1 COLLISION FOUND WITH %s !"),
 679                             sha1_to_hex(data->entry->idx.sha1));
 680                 size -= len;
 681                 buf += len;
 682         }
 683         return 0;
 684 }
 685
 686 static int check_collison(struct object_entry *entry)
 687 {
 688         struct compare_data data;
 689         enum object_type type;
 690         unsigned long size;
 691
 692         if (entry->size <= big_file_threshold || entry->type != OBJ_BLOB)
 693                 return -1;
 694
 695         memset(&data, 0, sizeof(data));
 696         data.entry = entry;
 697         data.st = open_istream(entry->idx.sha1, &type, &size, NULL);
 698         if (!data.st)
 699                 return -1;
 700         if (size != entry->size || type != entry->type)
 701                 die(_("SHA1 COLLISION FOUND WITH %s !"),
 702                     sha1_to_hex(entry->idx.sha1));
 703         unpack_data(entry, compare_objects, &data);
 704         close_istream(data.st);
 705         free(data.buf);
 706         return 0;
 707 }
 708
 709 static void sha1_object(const void *data, struct object_entry *obj_entry,
 710                         unsigned long size, enum object_type type,
 711                         const unsigned char *sha1)
 712 {
 713         void *new_data = NULL;
 714         int collision_test_needed;
 715
 716         assert(data || obj_entry);
 717
 718         read_lock();
 719         collision_test_needed = has_sha1_file(sha1);
 720         read_unlock();
 721
 722         if (collision_test_needed && !data) {
 723                 read_lock();
 724                 if (!check_collison(obj_entry))
 725                         collision_test_needed = 0;
 726                 read_unlock();
 727         }
 728         if (collision_test_needed) {
 729                 void *has_data;
 730                 enum object_type has_type;
 731                 unsigned long has_size;
 732                 read_lock();
 733                 has_type = sha1_object_info(sha1, &has_size);
 734                 if (has_type != type || has_size != size)
 735                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 736                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 737                 read_unlock();
 738                 if (!data)
 739                         data = new_data = get_data_from_pack(obj_entry);
 740                 if (!has_data)
 741                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 742                 if (size != has_size || type != has_type ||
 743                     memcmp(data, has_data, size) != 0)
 744                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 745                 free(has_data);
 746         }
 747
 748         if (strict) {
 749                 read_lock();
 750                 if (type == OBJ_BLOB) {
 751                         struct blob *blob = lookup_blob(sha1);
 752                         if (blob)
 753                                 blob->object.flags |= FLAG_CHECKED;
 754                         else
 755                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 756                 } else {
 757                         struct object *obj;
 758                         int eaten;
 759                         void *buf = (void *) data;
 760
 761                         assert(data && "data can only be NULL for large _blobs_");
 762
 763                         /*
 764                          * we do not need to free the memory here, as the
 765                          * buf is deleted by the caller.
 766                          */
 767                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 768                         if (!obj)
 769                                 die(_("invalid %s"), typename(type));
 770                         if (do_fsck_object &&
 771                             fsck_object(obj, 1, fsck_error_function))
 772                                 die(_("Error in object"));
 773                         if (fsck_walk(obj, mark_link, NULL))
 774                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 775
 776                         if (obj->type == OBJ_TREE) {
 777                                 struct tree *item = (struct tree *) obj;
 778                                 item->buffer = NULL;
 779                                 obj->parsed = 0;
 780                         }
 781                         if (obj->type == OBJ_COMMIT) {
 782                                 struct commit *commit = (struct commit *) obj;
 783                                 commit->buffer = NULL;
 784                         }
 785                         obj->flags |= FLAG_CHECKED;
 786                 }
 787                 read_unlock();
 788         }
 789
 790         free(new_data);
 791 }
 792
 793 /*
 794  * This function is part of find_unresolved_deltas(). There are two
 795  * walkers going in the opposite ways.
 796  *
 797  * The first one in find_unresolved_deltas() traverses down from
 798  * parent node to children, deflating nodes along the way. However,
 799  * memory for deflated nodes is limited by delta_base_cache_limit, so
 800  * at some point parent node's deflated content may be freed.
 801  *
 802  * The second walker is this function, which goes from current node up
 803  * to top parent if necessary to deflate the node. In normal
 804  * situation, its parent node would be already deflated, so it just
 805  * needs to apply delta.
 806  *
 807  * In the worst case scenario, parent node is no longer deflated because
 808  * we're running out of delta_base_cache_limit; we need to re-deflate
 809  * parents, possibly up to the top base.
 810  *
 811  * All deflated objects here are subject to be freed if we exceed
 812  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 813  * just need to make sure the last node is not freed.
 814  */
 815 static void *get_base_data(struct base_data *c)
 816 {
 817         if (!c->data) {
 818                 struct object_entry *obj = c->obj;
 819                 struct base_data **delta = NULL;
 820                 int delta_nr = 0, delta_alloc = 0;
 821
 822                 while (is_delta_type(c->obj->type) && !c->data) {
 823                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 824                         delta[delta_nr++] = c;
 825                         c = c->base;
 826                 }
 827                 if (!delta_nr) {
 828                         c->data = get_data_from_pack(obj);
 829                         c->size = obj->size;
 830                         get_thread_data()->base_cache_used += c->size;
 831                         prune_base_data(c);
 832                 }
 833                 for (; delta_nr > 0; delta_nr--) {
 834                         void *base, *raw;
 835                         c = delta[delta_nr - 1];
 836                         obj = c->obj;
 837                         base = get_base_data(c->base);
 838                         raw = get_data_from_pack(obj);
 839                         c->data = patch_delta(
 840                                 base, c->base->size,
 841                                 raw, obj->size,
 842                                 &c->size);
 843                         free(raw);
 844                         if (!c->data)
 845                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 846                         get_thread_data()->base_cache_used += c->size;
 847                         prune_base_data(c);
 848                 }
 849                 free(delta);
 850         }
 851         return c->data;
 852 }
 853
 854 static void resolve_delta(struct object_entry *delta_obj,
 855                           struct base_data *base, struct base_data *result)
 856 {
 857         void *base_data, *delta_data;
 858
 859         if (show_stat) {
 860                 delta_obj->delta_depth = base->obj->delta_depth + 1;
 861                 deepest_delta_lock();
 862                 if (deepest_delta < delta_obj->delta_depth)
 863                         deepest_delta = delta_obj->delta_depth;
 864                 deepest_delta_unlock();
 865         }
 866         delta_obj->base_object_no = base->obj - objects;
 867         delta_data = get_data_from_pack(delta_obj);
 868         base_data = get_base_data(base);
 869         result->obj = delta_obj;
 870         result->data = patch_delta(base_data, base->size,
 871                                    delta_data, delta_obj->size, &result->size);
 872         free(delta_data);
 873         if (!result->data)
 874                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 875         hash_sha1_file(result->data, result->size,
 876                        typename(delta_obj->real_type), delta_obj->idx.sha1);
 877         sha1_object(result->data, NULL, result->size, delta_obj->real_type,
 878                     delta_obj->idx.sha1);
 879         counter_lock();
 880         nr_resolved_deltas++;
 881         counter_unlock();
 882 }
 883
 884 /*
 885  * Standard boolean compare-and-swap: atomically check whether "*type" is
 886  * "want"; if so, swap in "set" and return true. Otherwise, leave it untouched
 887  * and return false.
 888  */
 889 static int compare_and_swap_type(enum object_type *type,
 890                                  enum object_type want,
 891                                  enum object_type set)
 892 {
 893         enum object_type old;
 894
 895         type_cas_lock();
 896         old = *type;
 897         if (old == want)
 898                 *type = set;
 899         type_cas_unlock();
 900
 901         return old == want;
 902 }
 903
 904 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 905                                                   struct base_data *prev_base)
 906 {
 907         if (base->ref_last == -1 && base->ofs_last == -1) {
 908                 union delta_base base_spec;
 909
 910                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 911                 find_delta_children(&base_spec,
 912                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 913
 914                 memset(&base_spec, 0, sizeof(base_spec));
 915                 base_spec.offset = base->obj->idx.offset;
 916                 find_delta_children(&base_spec,
 917                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 918
 919                 if (base->ref_last == -1 && base->ofs_last == -1) {
 920                         free(base->data);
 921                         return NULL;
 922                 }
 923
 924                 link_base_data(prev_base, base);
 925         }
 926
 927         if (base->ref_first <= base->ref_last) {
 928                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 929                 struct base_data *result = alloc_base_data();
 930
 931                 if (!compare_and_swap_type(&child->real_type, OBJ_REF_DELTA,
 932                                            base->obj->real_type))
 933                         die("BUG: child->real_type != OBJ_REF_DELTA");
 934
 935                 resolve_delta(child, base, result);
 936                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 937                         free_base_data(base);
 938
 939                 base->ref_first++;
 940                 return result;
 941         }
 942
 943         if (base->ofs_first <= base->ofs_last) {
 944                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 945                 struct base_data *result = alloc_base_data();
 946
 947                 assert(child->real_type == OBJ_OFS_DELTA);
 948                 child->real_type = base->obj->real_type;
 949                 resolve_delta(child, base, result);
 950                 if (base->ofs_first == base->ofs_last)
 951                         free_base_data(base);
 952
 953                 base->ofs_first++;
 954                 return result;
 955         }
 956
 957         unlink_base_data(base);
 958         return NULL;
 959 }
 960
 961 static void find_unresolved_deltas(struct base_data *base)
 962 {
 963         struct base_data *new_base, *prev_base = NULL;
 964         for (;;) {
 965                 new_base = find_unresolved_deltas_1(base, prev_base);
 966
 967                 if (new_base) {
 968                         prev_base = base;
 969                         base = new_base;
 970                 } else {
 971                         free(base);
 972                         base = prev_base;
 973                         if (!base)
 974                                 return;
 975                         prev_base = base->base;
 976                 }
 977         }
 978 }
 979
 980 static int compare_delta_entry(const void *a, const void *b)
 981 {
 982         const struct delta_entry *delta_a = a;
 983         const struct delta_entry *delta_b = b;
 984
 985         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 986         return compare_delta_bases(&delta_a->base, &delta_b->base,
 987                                    objects[delta_a->obj_no].type,
 988                                    objects[delta_b->obj_no].type);
 989 }
 990
 991 static void resolve_base(struct object_entry *obj)
 992 {
 993         struct base_data *base_obj = alloc_base_data();
 994         base_obj->obj = obj;
 995         base_obj->data = NULL;
 996         find_unresolved_deltas(base_obj);
 997 }
 998
 999 #ifndef NO_PTHREADS
1000 static void *threaded_second_pass(void *data)
1001 {
1002         set_thread_data(data);
1003         for (;;) {
1004                 int i;
1005                 counter_lock();
1006                 display_progress(progress, nr_resolved_deltas);
1007                 counter_unlock();
1008                 work_lock();
1009                 while (nr_dispatched < nr_objects &&
1010                        is_delta_type(objects[nr_dispatched].type))
1011                         nr_dispatched++;
1012                 if (nr_dispatched >= nr_objects) {
1013                         work_unlock();
1014                         break;
1015                 }
1016                 i = nr_dispatched++;
1017                 work_unlock();
1018
1019                 resolve_base(&objects[i]);
1020         }
1021         return NULL;
1022 }
1023 #endif
1024
1025 /*
1026  * First pass:
1027  * - find locations of all objects;
1028  * - calculate SHA1 of all non-delta objects;
1029  * - remember base (SHA1 or offset) for all deltas.
1030  */
1031 static void parse_pack_objects(unsigned char *sha1)
1032 {
1033         int i, nr_delays = 0;
1034         struct delta_entry *delta = deltas;
1035         struct stat st;
1036
1037         if (verbose)
1038                 progress = start_progress(
1039                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
1040                                 nr_objects);
1041         for (i = 0; i < nr_objects; i++) {
1042                 struct object_entry *obj = &objects[i];
1043                 void *data = unpack_raw_entry(obj, &delta->base, obj->idx.sha1);
1044                 obj->real_type = obj->type;
1045                 if (is_delta_type(obj->type)) {
1046                         nr_deltas++;
1047                         delta->obj_no = i;
1048                         delta++;
1049                 } else if (!data) {
1050                         /* large blobs, check later */
1051                         obj->real_type = OBJ_BAD;
1052                         nr_delays++;
1053                 } else
1054                         sha1_object(data, NULL, obj->size, obj->type, obj->idx.sha1);
1055                 free(data);
1056                 display_progress(progress, i+1);
1057         }
1058         objects[i].idx.offset = consumed_bytes;
1059         stop_progress(&progress);
1060
1061         /* Check pack integrity */
1062         flush();
1063         git_SHA1_Final(sha1, &input_ctx);
1064         if (hashcmp(fill(20), sha1))
1065                 die(_("pack is corrupted (SHA1 mismatch)"));
1066         use(20);
1067
1068         /* If input_fd is a file, we should have reached its end now. */
1069         if (fstat(input_fd, &st))
1070                 die_errno(_("cannot fstat packfile"));
1071         if (S_ISREG(st.st_mode) &&
1072                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
1073                 die(_("pack has junk at the end"));
1074
1075         for (i = 0; i < nr_objects; i++) {
1076                 struct object_entry *obj = &objects[i];
1077                 if (obj->real_type != OBJ_BAD)
1078                         continue;
1079                 obj->real_type = obj->type;
1080                 sha1_object(NULL, obj, obj->size, obj->type, obj->idx.sha1);
1081                 nr_delays--;
1082         }
1083         if (nr_delays)
1084                 die(_("confusion beyond insanity in parse_pack_objects()"));
1085 }
1086
1087 /*
1088  * Second pass:
1089  * - for all non-delta objects, look if it is used as a base for
1090  *   deltas;
1091  * - if used as a base, uncompress the object and apply all deltas,
1092  *   recursively checking if the resulting object is used as a base
1093  *   for some more deltas.
1094  */
1095 static void resolve_deltas(void)
1096 {
1097         int i;
1098
1099         if (!nr_deltas)
1100                 return;
1101
1102         /* Sort deltas by base SHA1/offset for fast searching */
1103         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
1104               compare_delta_entry);
1105
1106         if (verbose)
1107                 progress = start_progress(_("Resolving deltas"), nr_deltas);
1108
1109 #ifndef NO_PTHREADS
1110         nr_dispatched = 0;
1111         if (nr_threads > 1 || getenv("GIT_FORCE_THREADS")) {
1112                 init_thread();
1113                 for (i = 0; i < nr_threads; i++) {
1114                         int ret = pthread_create(&thread_data[i].thread, NULL,
1115                                                  threaded_second_pass, thread_data + i);
1116                         if (ret)
1117                                 die(_("unable to create thread: %s"),
1118                                     strerror(ret));
1119                 }
1120                 for (i = 0; i < nr_threads; i++)
1121                         pthread_join(thread_data[i].thread, NULL);
1122                 cleanup_thread();
1123                 return;
1124         }
1125 #endif
1126
1127         for (i = 0; i < nr_objects; i++) {
1128                 struct object_entry *obj = &objects[i];
1129
1130                 if (is_delta_type(obj->type))
1131                         continue;
1132                 resolve_base(obj);
1133                 display_progress(progress, nr_resolved_deltas);
1134         }
1135 }
1136
1137 /*
1138  * Third pass:
1139  * - append objects to convert thin pack to full pack if required
1140  * - write the final 20-byte SHA-1
1141  */
1142 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved);
1143 static void conclude_pack(int fix_thin_pack, const char *curr_pack, unsigned char *pack_sha1)
1144 {
1145         if (nr_deltas == nr_resolved_deltas) {
1146                 stop_progress(&progress);
1147                 /* Flush remaining pack final 20-byte SHA1. */
1148                 flush();
1149                 return;
1150         }
1151
1152         if (fix_thin_pack) {
1153                 struct sha1file *f;
1154                 unsigned char read_sha1[20], tail_sha1[20];
1155                 struct strbuf msg = STRBUF_INIT;
1156                 int nr_unresolved = nr_deltas - nr_resolved_deltas;
1157                 int nr_objects_initial = nr_objects;
1158                 if (nr_unresolved <= 0)
1159                         die(_("confusion beyond insanity"));
1160                 objects = xrealloc(objects,
1161                                    (nr_objects + nr_unresolved + 1)
1162                                    * sizeof(*objects));
1163                 memset(objects + nr_objects + 1, 0,
1164                        nr_unresolved * sizeof(*objects));
1165                 f = sha1fd(output_fd, curr_pack);
1166                 fix_unresolved_deltas(f, nr_unresolved);
1167                 strbuf_addf(&msg, _("completed with %d local objects"),
1168                             nr_objects - nr_objects_initial);
1169                 stop_progress_msg(&progress, msg.buf);
1170                 strbuf_release(&msg);
1171                 sha1close(f, tail_sha1, 0);
1172                 hashcpy(read_sha1, pack_sha1);
1173                 fixup_pack_header_footer(output_fd, pack_sha1,
1174                                          curr_pack, nr_objects,
1175                                          read_sha1, consumed_bytes-20);
1176                 if (hashcmp(read_sha1, tail_sha1) != 0)
1177                         die(_("Unexpected tail checksum for %s "
1178                               "(disk corruption?)"), curr_pack);
1179         }
1180         if (nr_deltas != nr_resolved_deltas)
1181                 die(Q_("pack has %d unresolved delta",
1182                        "pack has %d unresolved deltas",
1183                        nr_deltas - nr_resolved_deltas),
1184                     nr_deltas - nr_resolved_deltas);
1185 }
1186
1187 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
1188 {
1189         git_zstream stream;
1190         int status;
1191         unsigned char outbuf[4096];
1192
1193         memset(&stream, 0, sizeof(stream));
1194         git_deflate_init(&stream, zlib_compression_level);
1195         stream.next_in = in;
1196         stream.avail_in = size;
1197
1198         do {
1199                 stream.next_out = outbuf;
1200                 stream.avail_out = sizeof(outbuf);
1201                 status = git_deflate(&stream, Z_FINISH);
1202                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
1203         } while (status == Z_OK);
1204
1205         if (status != Z_STREAM_END)
1206                 die(_("unable to deflate appended object (%d)"), status);
1207         size = stream.total_out;
1208         git_deflate_end(&stream);
1209         return size;
1210 }
1211
1212 static struct object_entry *append_obj_to_pack(struct sha1file *f,
1213                                const unsigned char *sha1, void *buf,
1214                                unsigned long size, enum object_type type)
1215 {
1216         struct object_entry *obj = &objects[nr_objects++];
1217         unsigned char header[10];
1218         unsigned long s = size;
1219         int n = 0;
1220         unsigned char c = (type << 4) | (s & 15);
1221         s >>= 4;
1222         while (s) {
1223                 header[n++] = c | 0x80;
1224                 c = s & 0x7f;
1225                 s >>= 7;
1226         }
1227         header[n++] = c;
1228         crc32_begin(f);
1229         sha1write(f, header, n);
1230         obj[0].size = size;
1231         obj[0].hdr_size = n;
1232         obj[0].type = type;
1233         obj[0].real_type = type;
1234         obj[1].idx.offset = obj[0].idx.offset + n;
1235         obj[1].idx.offset += write_compressed(f, buf, size);
1236         obj[0].idx.crc32 = crc32_end(f);
1237         sha1flush(f);
1238         hashcpy(obj->idx.sha1, sha1);
1239         return obj;
1240 }
1241
1242 static int delta_pos_compare(const void *_a, const void *_b)
1243 {
1244         struct delta_entry *a = *(struct delta_entry **)_a;
1245         struct delta_entry *b = *(struct delta_entry **)_b;
1246         return a->obj_no - b->obj_no;
1247 }
1248
1249 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
1250 {
1251         struct delta_entry **sorted_by_pos;
1252         int i, n = 0;
1253
1254         /*
1255          * Since many unresolved deltas may well be themselves base objects
1256          * for more unresolved deltas, we really want to include the
1257          * smallest number of base objects that would cover as much delta
1258          * as possible by picking the
1259          * trunc deltas first, allowing for other deltas to resolve without
1260          * additional base objects.  Since most base objects are to be found
1261          * before deltas depending on them, a good heuristic is to start
1262          * resolving deltas in the same order as their position in the pack.
1263          */
1264         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
1265         for (i = 0; i < nr_deltas; i++) {
1266                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
1267                         continue;
1268                 sorted_by_pos[n++] = &deltas[i];
1269         }
1270         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
1271
1272         for (i = 0; i < n; i++) {
1273                 struct delta_entry *d = sorted_by_pos[i];
1274                 enum object_type type;
1275                 struct base_data *base_obj = alloc_base_data();
1276
1277                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
1278                         continue;
1279                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
1280                 if (!base_obj->data)
1281                         continue;
1282
1283                 if (check_sha1_signature(d->base.sha1, base_obj->data,
1284                                 base_obj->size, typename(type)))
1285                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
1286                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
1287                                         base_obj->data, base_obj->size, type);
1288                 find_unresolved_deltas(base_obj);
1289                 display_progress(progress, nr_resolved_deltas);
1290         }
1291         free(sorted_by_pos);
1292 }
1293
1294 static void final(const char *final_pack_name, const char *curr_pack_name,
1295                   const char *final_index_name, const char *curr_index_name,
1296                   const char *keep_name, const char *keep_msg,
1297                   unsigned char *sha1)
1298 {
1299         const char *report = "pack";
1300         char name[PATH_MAX];
1301         int err;
1302
1303         if (!from_stdin) {
1304                 close(input_fd);
1305         } else {
1306                 fsync_or_die(output_fd, curr_pack_name);
1307                 err = close(output_fd);
1308                 if (err)
1309                         die_errno(_("error while closing pack file"));
1310         }
1311
1312         if (keep_msg) {
1313                 int keep_fd, keep_msg_len = strlen(keep_msg);
1314
1315                 if (!keep_name)
1316                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
1317                 else
1318                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
1319
1320                 if (keep_fd < 0) {
1321                         if (errno != EEXIST)
1322                                 die_errno(_("cannot write keep file '%s'"),
1323                                           keep_name);
1324                 } else {
1325                         if (keep_msg_len > 0) {
1326                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
1327                                 write_or_die(keep_fd, "\n", 1);
1328                         }
1329                         if (close(keep_fd) != 0)
1330                                 die_errno(_("cannot close written keep file '%s'"),
1331                                     keep_name);
1332                         report = "keep";
1333                 }
1334         }
1335
1336         if (final_pack_name != curr_pack_name) {
1337                 if (!final_pack_name) {
1338                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
1339                                  get_object_directory(), sha1_to_hex(sha1));
1340                         final_pack_name = name;
1341                 }
1342                 if (move_temp_to_file(curr_pack_name, final_pack_name))
1343                         die(_("cannot store pack file"));
1344         } else if (from_stdin)
1345                 chmod(final_pack_name, 0444);
1346
1347         if (final_index_name != curr_index_name) {
1348                 if (!final_index_name) {
1349                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
1350                                  get_object_directory(), sha1_to_hex(sha1));
1351                         final_index_name = name;
1352                 }
1353                 if (move_temp_to_file(curr_index_name, final_index_name))
1354                         die(_("cannot store index file"));
1355         } else
1356                 chmod(final_index_name, 0444);
1357
1358         if (!from_stdin) {
1359                 printf("%s\n", sha1_to_hex(sha1));
1360         } else {
1361                 char buf[48];
1362                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
1363                                    report, sha1_to_hex(sha1));
1364                 write_or_die(1, buf, len);
1365
1366                 /*
1367                  * Let's just mimic git-unpack-objects here and write
1368                  * the last part of the input buffer to stdout.
1369                  */
1370                 while (input_len) {
1371                         err = xwrite(1, input_buffer + input_offset, input_len);
1372                         if (err <= 0)
1373                                 break;
1374                         input_len -= err;
1375                         input_offset += err;
1376                 }
1377         }
1378 }
1379
1380 static int git_index_pack_config(const char *k, const char *v, void *cb)
1381 {
1382         struct pack_idx_option *opts = cb;
1383
1384         if (!strcmp(k, "pack.indexversion")) {
1385                 opts->version = git_config_int(k, v);
1386                 if (opts->version > 2)
1387                         die(_("bad pack.indexversion=%"PRIu32), opts->version);
1388                 return 0;
1389         }
1390         if (!strcmp(k, "pack.threads")) {
1391                 nr_threads = git_config_int(k, v);
1392                 if (nr_threads < 0)
1393                         die(_("invalid number of threads specified (%d)"),
1394                             nr_threads);
1395 #ifdef NO_PTHREADS
1396                 if (nr_threads != 1)
1397                         warning(_("no threads support, ignoring %s"), k);
1398                 nr_threads = 1;
1399 #endif
1400                 return 0;
1401         }
1402         return git_default_config(k, v, cb);
1403 }
1404
1405 static int cmp_uint32(const void *a_, const void *b_)
1406 {
1407         uint32_t a = *((uint32_t *)a_);
1408         uint32_t b = *((uint32_t *)b_);
1409
1410         return (a < b) ? -1 : (a != b);
1411 }
1412
1413 static void read_v2_anomalous_offsets(struct packed_git *p,
1414                                       struct pack_idx_option *opts)
1415 {
1416         const uint32_t *idx1, *idx2;
1417         uint32_t i;
1418
1419         /* The address of the 4-byte offset table */
1420         idx1 = (((const uint32_t *)p->index_data)
1421                 + 2 /* 8-byte header */
1422                 + 256 /* fan out */
1423                 + 5 * p->num_objects /* 20-byte SHA-1 table */
1424                 + p->num_objects /* CRC32 table */
1425                 );
1426
1427         /* The address of the 8-byte offset table */
1428         idx2 = idx1 + p->num_objects;
1429
1430         for (i = 0; i < p->num_objects; i++) {
1431                 uint32_t off = ntohl(idx1[i]);
1432                 if (!(off & 0x80000000))
1433                         continue;
1434                 off = off & 0x7fffffff;
1435                 if (idx2[off * 2])
1436                         continue;
1437                 /*
1438                  * The real offset is ntohl(idx2[off * 2]) in high 4
1439                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1440                  * octets.  But idx2[off * 2] is Zero!!!
1441                  */
1442                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1443                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1444         }
1445
1446         if (1 < opts->anomaly_nr)
1447                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1448 }
1449
1450 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1451 {
1452         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1453
1454         if (!p)
1455                 die(_("Cannot open existing pack file '%s'"), pack_name);
1456         if (open_pack_index(p))
1457                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1458
1459         /* Read the attributes from the existing idx file */
1460         opts->version = p->index_version;
1461
1462         if (opts->version == 2)
1463                 read_v2_anomalous_offsets(p, opts);
1464
1465         /*
1466          * Get rid of the idx file as we do not need it anymore.
1467          * NEEDSWORK: extract this bit from free_pack_by_name() in
1468          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1469          * know we haven't installed this pack (hence we never have
1470          * read anything from it).
1471          */
1472         close_pack_index(p);
1473         free(p);
1474 }
1475
1476 static void show_pack_info(int stat_only)
1477 {
1478         int i, baseobjects = nr_objects - nr_deltas;
1479         unsigned long *chain_histogram = NULL;
1480
1481         if (deepest_delta)
1482                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1483
1484         for (i = 0; i < nr_objects; i++) {
1485                 struct object_entry *obj = &objects[i];
1486
1487                 if (is_delta_type(obj->type))
1488                         chain_histogram[obj->delta_depth - 1]++;
1489                 if (stat_only)
1490                         continue;
1491                 printf("%s %-6s %lu %lu %"PRIuMAX,
1492                        sha1_to_hex(obj->idx.sha1),
1493                        typename(obj->real_type), obj->size,
1494                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1495                        (uintmax_t)obj->idx.offset);
1496                 if (is_delta_type(obj->type)) {
1497                         struct object_entry *bobj = &objects[obj->base_object_no];
1498                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1499                 }
1500                 putchar('\n');
1501         }
1502
1503         if (baseobjects)
1504                 printf_ln(Q_("non delta: %d object",
1505                              "non delta: %d objects",
1506                              baseobjects),
1507                           baseobjects);
1508         for (i = 0; i < deepest_delta; i++) {
1509                 if (!chain_histogram[i])
1510                         continue;
1511                 printf_ln(Q_("chain length = %d: %lu object",
1512                              "chain length = %d: %lu objects",
1513                              chain_histogram[i]),
1514                           i + 1,
1515                           chain_histogram[i]);
1516         }
1517 }
1518
1519 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1520 {
1521         int i, fix_thin_pack = 0, verify = 0, stat_only = 0;
1522         const char *curr_pack, *curr_index;
1523         const char *index_name = NULL, *pack_name = NULL;
1524         const char *keep_name = NULL, *keep_msg = NULL;
1525         char *index_name_buf = NULL, *keep_name_buf = NULL;
1526         struct pack_idx_entry **idx_objects;
1527         struct pack_idx_option opts;
1528         unsigned char pack_sha1[20];
1529         unsigned foreign_nr = 1;        /* zero is a "good" value, assume bad */
1530
1531         if (argc == 2 && !strcmp(argv[1], "-h"))
1532                 usage(index_pack_usage);
1533
1534         read_replace_refs = 0;
1535
1536         reset_pack_idx_option(&opts);
1537         git_config(git_index_pack_config, &opts);
1538         if (prefix && chdir(prefix))
1539                 die(_("Cannot come back to cwd"));
1540
1541         for (i = 1; i < argc; i++) {
1542                 const char *arg = argv[i];
1543
1544                 if (*arg == '-') {
1545                         if (!strcmp(arg, "--stdin")) {
1546                                 from_stdin = 1;
1547                         } else if (!strcmp(arg, "--fix-thin")) {
1548                                 fix_thin_pack = 1;
1549                         } else if (!strcmp(arg, "--strict")) {
1550                                 strict = 1;
1551                                 do_fsck_object = 1;
1552                         } else if (!strcmp(arg, "--check-self-contained-and-connected")) {
1553                                 strict = 1;
1554                                 check_self_contained_and_connected = 1;
1555                         } else if (!strcmp(arg, "--verify")) {
1556                                 verify = 1;
1557                         } else if (!strcmp(arg, "--verify-stat")) {
1558                                 verify = 1;
1559                                 show_stat = 1;
1560                         } else if (!strcmp(arg, "--verify-stat-only")) {
1561                                 verify = 1;
1562                                 show_stat = 1;
1563                                 stat_only = 1;
1564                         } else if (!strcmp(arg, "--keep")) {
1565                                 keep_msg = "";
1566                         } else if (!prefixcmp(arg, "--keep=")) {
1567                                 keep_msg = arg + 7;
1568                         } else if (!prefixcmp(arg, "--threads=")) {
1569                                 char *end;
1570                                 nr_threads = strtoul(arg+10, &end, 0);
1571                                 if (!arg[10] || *end || nr_threads < 0)
1572                                         usage(index_pack_usage);
1573 #ifdef NO_PTHREADS
1574                                 if (nr_threads != 1)
1575                                         warning(_("no threads support, "
1576                                                   "ignoring %s"), arg);
1577                                 nr_threads = 1;
1578 #endif
1579                         } else if (!prefixcmp(arg, "--pack_header=")) {
1580                                 struct pack_header *hdr;
1581                                 char *c;
1582
1583                                 hdr = (struct pack_header *)input_buffer;
1584                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1585                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1586                                 if (*c != ',')
1587                                         die(_("bad %s"), arg);
1588                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1589                                 if (*c)
1590                                         die(_("bad %s"), arg);
1591                                 input_len = sizeof(*hdr);
1592                         } else if (!strcmp(arg, "-v")) {
1593                                 verbose = 1;
1594                         } else if (!strcmp(arg, "-o")) {
1595                                 if (index_name || (i+1) >= argc)
1596                                         usage(index_pack_usage);
1597                                 index_name = argv[++i];
1598                         } else if (!prefixcmp(arg, "--index-version=")) {
1599                                 char *c;
1600                                 opts.version = strtoul(arg + 16, &c, 10);
1601                                 if (opts.version > 2)
1602                                         die(_("bad %s"), arg);
1603                                 if (*c == ',')
1604                                         opts.off32_limit = strtoul(c+1, &c, 0);
1605                                 if (*c || opts.off32_limit & 0x80000000)
1606                                         die(_("bad %s"), arg);
1607                         } else
1608                                 usage(index_pack_usage);
1609                         continue;
1610                 }
1611
1612                 if (pack_name)
1613                         usage(index_pack_usage);
1614                 pack_name = arg;
1615         }
1616
1617         if (!pack_name && !from_stdin)
1618                 usage(index_pack_usage);
1619         if (fix_thin_pack && !from_stdin)
1620                 die(_("--fix-thin cannot be used without --stdin"));
1621         if (!index_name && pack_name) {
1622                 int len = strlen(pack_name);
1623                 if (!has_extension(pack_name, ".pack"))
1624                         die(_("packfile name '%s' does not end with '.pack'"),
1625                             pack_name);
1626                 index_name_buf = xmalloc(len);
1627                 memcpy(index_name_buf, pack_name, len - 5);
1628                 strcpy(index_name_buf + len - 5, ".idx");
1629                 index_name = index_name_buf;
1630         }
1631         if (keep_msg && !keep_name && pack_name) {
1632                 int len = strlen(pack_name);
1633                 if (!has_extension(pack_name, ".pack"))
1634                         die(_("packfile name '%s' does not end with '.pack'"),
1635                             pack_name);
1636                 keep_name_buf = xmalloc(len);
1637                 memcpy(keep_name_buf, pack_name, len - 5);
1638                 strcpy(keep_name_buf + len - 5, ".keep");
1639                 keep_name = keep_name_buf;
1640         }
1641         if (verify) {
1642                 if (!index_name)
1643                         die(_("--verify with no packfile name given"));
1644                 read_idx_option(&opts, index_name);
1645                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1646         }
1647         if (strict)
1648                 opts.flags |= WRITE_IDX_STRICT;
1649
1650 #ifndef NO_PTHREADS
1651         if (!nr_threads) {
1652                 nr_threads = online_cpus();
1653                 /* An experiment showed that more threads does not mean faster */
1654                 if (nr_threads > 3)
1655                         nr_threads = 3;
1656         }
1657 #endif
1658
1659         curr_pack = open_pack_file(pack_name);
1660         parse_pack_header();
1661         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1662         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1663         parse_pack_objects(pack_sha1);
1664         resolve_deltas();
1665         conclude_pack(fix_thin_pack, curr_pack, pack_sha1);
1666         free(deltas);
1667         if (strict)
1668                 foreign_nr = check_objects();
1669
1670         if (show_stat)
1671                 show_pack_info(stat_only);
1672
1673         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1674         for (i = 0; i < nr_objects; i++)
1675                 idx_objects[i] = &objects[i].idx;
1676         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1677         free(idx_objects);
1678
1679         if (!verify)
1680                 final(pack_name, curr_pack,
1681                       index_name, curr_index,
1682                       keep_name, keep_msg,
1683                       pack_sha1);
1684         else
1685                 close(input_fd);
1686         free(objects);
1687         free(index_name_buf);
1688         free(keep_name_buf);
1689         if (pack_name == NULL)
1690                 free((void *) curr_pack);
1691         if (index_name == NULL)
1692                 free((void *) curr_index);
1693
1694         /*
1695          * Let the caller know this pack is not self contained
1696          */
1697         if (check_self_contained_and_connected && foreign_nr)
1698                 return 1;
1699
1700         return 0;
1701 }