builtin/index-pack.c

   1 #include "cache.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
  13 static const char index_pack_usage[] =
  14 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  15
  16 struct object_entry
  17 {
  18         struct pack_idx_entry idx;
  19         unsigned long size;
  20         unsigned int hdr_size;
  21         enum object_type type;
  22         enum object_type real_type;
  23 };
  24
  25 union delta_base {
  26         unsigned char sha1[20];
  27         off_t offset;
  28 };
  29
  30 struct base_data {
  31         struct base_data *base;
  32         struct base_data *child;
  33         struct object_entry *obj;
  34         void *data;
  35         unsigned long size;
  36 };
  37
  38 /*
  39  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  40  * to memcmp() only the first 20 bytes.
  41  */
  42 #define UNION_BASE_SZ   20
  43
  44 #define FLAG_LINK (1u<<20)
  45 #define FLAG_CHECKED (1u<<21)
  46
  47 struct delta_entry
  48 {
  49         union delta_base base;
  50         int obj_no;
  51 };
  52
  53 static struct object_entry *objects;
  54 static struct delta_entry *deltas;
  55 static struct base_data *base_cache;
  56 static size_t base_cache_used;
  57 static int nr_objects;
  58 static int nr_deltas;
  59 static int nr_resolved_deltas;
  60
  61 static int from_stdin;
  62 static int strict;
  63 static int verbose;
  64
  65 static struct progress *progress;
  66
  67 /* We always read in 4kB chunks. */
  68 static unsigned char input_buffer[4096];
  69 static unsigned int input_offset, input_len;
  70 static off_t consumed_bytes;
  71 static git_SHA_CTX input_ctx;
  72 static uint32_t input_crc32;
  73 static int input_fd, output_fd, pack_fd;
  74
  75 static int mark_link(struct object *obj, int type, void *data)
  76 {
  77         if (!obj)
  78                 return -1;
  79
  80         if (type != OBJ_ANY && obj->type != type)
  81                 die("object type mismatch at %s", sha1_to_hex(obj->sha1));
  82
  83         obj->flags |= FLAG_LINK;
  84         return 0;
  85 }
  86
  87 /* The content of each linked object must have been checked
  88    or it must be already present in the object database */
  89 static void check_object(struct object *obj)
  90 {
  91         if (!obj)
  92                 return;
  93
  94         if (!(obj->flags & FLAG_LINK))
  95                 return;
  96
  97         if (!(obj->flags & FLAG_CHECKED)) {
  98                 unsigned long size;
  99                 int type = sha1_object_info(obj->sha1, &size);
 100                 if (type != obj->type || type <= 0)
 101                         die("object of unexpected type");
 102                 obj->flags |= FLAG_CHECKED;
 103                 return;
 104         }
 105 }
 106
 107 static void check_objects(void)
 108 {
 109         unsigned i, max;
 110
 111         max = get_max_object_index();
 112         for (i = 0; i < max; i++)
 113                 check_object(get_indexed_object(i));
 114 }
 115
 116
 117 /* Discard current buffer used content. */
 118 static void flush(void)
 119 {
 120         if (input_offset) {
 121                 if (output_fd >= 0)
 122                         write_or_die(output_fd, input_buffer, input_offset);
 123                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 124                 memmove(input_buffer, input_buffer + input_offset, input_len);
 125                 input_offset = 0;
 126         }
 127 }
 128
 129 /*
 130  * Make sure at least "min" bytes are available in the buffer, and
 131  * return the pointer to the buffer.
 132  */
 133 static void *fill(int min)
 134 {
 135         if (min <= input_len)
 136                 return input_buffer + input_offset;
 137         if (min > sizeof(input_buffer))
 138                 die("cannot fill %d bytes", min);
 139         flush();
 140         do {
 141                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 142                                 sizeof(input_buffer) - input_len);
 143                 if (ret <= 0) {
 144                         if (!ret)
 145                                 die("early EOF");
 146                         die_errno("read error on input");
 147                 }
 148                 input_len += ret;
 149                 if (from_stdin)
 150                         display_throughput(progress, consumed_bytes + input_len);
 151         } while (input_len < min);
 152         return input_buffer;
 153 }
 154
 155 static void use(int bytes)
 156 {
 157         if (bytes > input_len)
 158                 die("used more bytes than were available");
 159         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 160         input_len -= bytes;
 161         input_offset += bytes;
 162
 163         /* make sure off_t is sufficiently large not to wrap */
 164         if (signed_add_overflows(consumed_bytes, bytes))
 165                 die("pack too large for current definition of off_t");
 166         consumed_bytes += bytes;
 167 }
 168
 169 static const char *open_pack_file(const char *pack_name)
 170 {
 171         if (from_stdin) {
 172                 input_fd = 0;
 173                 if (!pack_name) {
 174                         static char tmpfile[PATH_MAX];
 175                         output_fd = odb_mkstemp(tmpfile, sizeof(tmpfile),
 176                                                 "pack/tmp_pack_XXXXXX");
 177                         pack_name = xstrdup(tmpfile);
 178                 } else
 179                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 180                 if (output_fd < 0)
 181                         die_errno("unable to create '%s'", pack_name);
 182                 pack_fd = output_fd;
 183         } else {
 184                 input_fd = open(pack_name, O_RDONLY);
 185                 if (input_fd < 0)
 186                         die_errno("cannot open packfile '%s'", pack_name);
 187                 output_fd = -1;
 188                 pack_fd = input_fd;
 189         }
 190         git_SHA1_Init(&input_ctx);
 191         return pack_name;
 192 }
 193
 194 static void parse_pack_header(void)
 195 {
 196         struct pack_header *hdr = fill(sizeof(struct pack_header));
 197
 198         /* Header consistency check */
 199         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 200                 die("pack signature mismatch");
 201         if (!pack_version_ok(hdr->hdr_version))
 202                 die("pack version %"PRIu32" unsupported",
 203                         ntohl(hdr->hdr_version));
 204
 205         nr_objects = ntohl(hdr->hdr_entries);
 206         use(sizeof(struct pack_header));
 207 }
 208
 209 static NORETURN void bad_object(unsigned long offset, const char *format,
 210                        ...) __attribute__((format (printf, 2, 3)));
 211
 212 static void bad_object(unsigned long offset, const char *format, ...)
 213 {
 214         va_list params;
 215         char buf[1024];
 216
 217         va_start(params, format);
 218         vsnprintf(buf, sizeof(buf), format, params);
 219         va_end(params);
 220         die("pack has bad object at offset %lu: %s", offset, buf);
 221 }
 222
 223 static void free_base_data(struct base_data *c)
 224 {
 225         if (c->data) {
 226                 free(c->data);
 227                 c->data = NULL;
 228                 base_cache_used -= c->size;
 229         }
 230 }
 231
 232 static void prune_base_data(struct base_data *retain)
 233 {
 234         struct base_data *b;
 235         for (b = base_cache;
 236              base_cache_used > delta_base_cache_limit && b;
 237              b = b->child) {
 238                 if (b->data && b != retain)
 239                         free_base_data(b);
 240         }
 241 }
 242
 243 static void link_base_data(struct base_data *base, struct base_data *c)
 244 {
 245         if (base)
 246                 base->child = c;
 247         else
 248                 base_cache = c;
 249
 250         c->base = base;
 251         c->child = NULL;
 252         if (c->data)
 253                 base_cache_used += c->size;
 254         prune_base_data(c);
 255 }
 256
 257 static void unlink_base_data(struct base_data *c)
 258 {
 259         struct base_data *base = c->base;
 260         if (base)
 261                 base->child = NULL;
 262         else
 263                 base_cache = NULL;
 264         free_base_data(c);
 265 }
 266
 267 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 268 {
 269         int status;
 270         z_stream stream;
 271         void *buf = xmalloc(size);
 272
 273         memset(&stream, 0, sizeof(stream));
 274         git_inflate_init(&stream);
 275         stream.next_out = buf;
 276         stream.avail_out = size;
 277
 278         do {
 279                 stream.next_in = fill(1);
 280                 stream.avail_in = input_len;
 281                 status = git_inflate(&stream, 0);
 282                 use(input_len - stream.avail_in);
 283         } while (status == Z_OK);
 284         if (stream.total_out != size || status != Z_STREAM_END)
 285                 bad_object(offset, "inflate returned %d", status);
 286         git_inflate_end(&stream);
 287         return buf;
 288 }
 289
 290 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 291 {
 292         unsigned char *p;
 293         unsigned long size, c;
 294         off_t base_offset;
 295         unsigned shift;
 296         void *data;
 297
 298         obj->idx.offset = consumed_bytes;
 299         input_crc32 = crc32(0, Z_NULL, 0);
 300
 301         p = fill(1);
 302         c = *p;
 303         use(1);
 304         obj->type = (c >> 4) & 7;
 305         size = (c & 15);
 306         shift = 4;
 307         while (c & 0x80) {
 308                 p = fill(1);
 309                 c = *p;
 310                 use(1);
 311                 size += (c & 0x7f) << shift;
 312                 shift += 7;
 313         }
 314         obj->size = size;
 315
 316         switch (obj->type) {
 317         case OBJ_REF_DELTA:
 318                 hashcpy(delta_base->sha1, fill(20));
 319                 use(20);
 320                 break;
 321         case OBJ_OFS_DELTA:
 322                 memset(delta_base, 0, sizeof(*delta_base));
 323                 p = fill(1);
 324                 c = *p;
 325                 use(1);
 326                 base_offset = c & 127;
 327                 while (c & 128) {
 328                         base_offset += 1;
 329                         if (!base_offset || MSB(base_offset, 7))
 330                                 bad_object(obj->idx.offset, "offset value overflow for delta base object");
 331                         p = fill(1);
 332                         c = *p;
 333                         use(1);
 334                         base_offset = (base_offset << 7) + (c & 127);
 335                 }
 336                 delta_base->offset = obj->idx.offset - base_offset;
 337                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 338                         bad_object(obj->idx.offset, "delta base offset is out of bound");
 339                 break;
 340         case OBJ_COMMIT:
 341         case OBJ_TREE:
 342         case OBJ_BLOB:
 343         case OBJ_TAG:
 344                 break;
 345         default:
 346                 bad_object(obj->idx.offset, "unknown object type %d", obj->type);
 347         }
 348         obj->hdr_size = consumed_bytes - obj->idx.offset;
 349
 350         data = unpack_entry_data(obj->idx.offset, obj->size);
 351         obj->idx.crc32 = input_crc32;
 352         return data;
 353 }
 354
 355 static void *get_data_from_pack(struct object_entry *obj)
 356 {
 357         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 358         unsigned long len = obj[1].idx.offset - from;
 359         unsigned char *data, *inbuf;
 360         z_stream stream;
 361         int status;
 362
 363         data = xmalloc(obj->size);
 364         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 365
 366         memset(&stream, 0, sizeof(stream));
 367         git_inflate_init(&stream);
 368         stream.next_out = data;
 369         stream.avail_out = obj->size;
 370
 371         do {
 372                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 373                 n = pread(pack_fd, inbuf, n, from);
 374                 if (n < 0)
 375                         die_errno("cannot pread pack file");
 376                 if (!n)
 377                         die("premature end of pack file, %lu bytes missing", len);
 378                 from += n;
 379                 len -= n;
 380                 stream.next_in = inbuf;
 381                 stream.avail_in = n;
 382                 status = git_inflate(&stream, 0);
 383         } while (len && status == Z_OK && !stream.avail_in);
 384
 385         /* This has been inflated OK when first encountered, so... */
 386         if (status != Z_STREAM_END || stream.total_out != obj->size)
 387                 die("serious inflate inconsistency");
 388
 389         git_inflate_end(&stream);
 390         free(inbuf);
 391         return data;
 392 }
 393
 394 static int compare_delta_bases(const union delta_base *base1,
 395                                const union delta_base *base2,
 396                                enum object_type type1,
 397                                enum object_type type2)
 398 {
 399         int cmp = type1 - type2;
 400         if (cmp)
 401                 return cmp;
 402         return memcmp(base1, base2, UNION_BASE_SZ);
 403 }
 404
 405 static int find_delta(const union delta_base *base, enum object_type type)
 406 {
 407         int first = 0, last = nr_deltas;
 408
 409         while (first < last) {
 410                 int next = (first + last) / 2;
 411                 struct delta_entry *delta = &deltas[next];
 412                 int cmp;
 413
 414                 cmp = compare_delta_bases(base, &delta->base,
 415                                           type, objects[delta->obj_no].type);
 416                 if (!cmp)
 417                         return next;
 418                 if (cmp < 0) {
 419                         last = next;
 420                         continue;
 421                 }
 422                 first = next+1;
 423         }
 424         return -first-1;
 425 }
 426
 427 static void find_delta_children(const union delta_base *base,
 428                                 int *first_index, int *last_index,
 429                                 enum object_type type)
 430 {
 431         int first = find_delta(base, type);
 432         int last = first;
 433         int end = nr_deltas - 1;
 434
 435         if (first < 0) {
 436                 *first_index = 0;
 437                 *last_index = -1;
 438                 return;
 439         }
 440         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 441                 --first;
 442         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 443                 ++last;
 444         *first_index = first;
 445         *last_index = last;
 446 }
 447
 448 static void sha1_object(const void *data, unsigned long size,
 449                         enum object_type type, unsigned char *sha1)
 450 {
 451         hash_sha1_file(data, size, typename(type), sha1);
 452         if (has_sha1_file(sha1)) {
 453                 void *has_data;
 454                 enum object_type has_type;
 455                 unsigned long has_size;
 456                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 457                 if (!has_data)
 458                         die("cannot read existing object %s", sha1_to_hex(sha1));
 459                 if (size != has_size || type != has_type ||
 460                     memcmp(data, has_data, size) != 0)
 461                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 462                 free(has_data);
 463         }
 464         if (strict) {
 465                 if (type == OBJ_BLOB) {
 466                         struct blob *blob = lookup_blob(sha1);
 467                         if (blob)
 468                                 blob->object.flags |= FLAG_CHECKED;
 469                         else
 470                                 die("invalid blob object %s", sha1_to_hex(sha1));
 471                 } else {
 472                         struct object *obj;
 473                         int eaten;
 474                         void *buf = (void *) data;
 475
 476                         /*
 477                          * we do not need to free the memory here, as the
 478                          * buf is deleted by the caller.
 479                          */
 480                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 481                         if (!obj)
 482                                 die("invalid %s", typename(type));
 483                         if (fsck_object(obj, 1, fsck_error_function))
 484                                 die("Error in object");
 485                         if (fsck_walk(obj, mark_link, NULL))
 486                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 487
 488                         if (obj->type == OBJ_TREE) {
 489                                 struct tree *item = (struct tree *) obj;
 490                                 item->buffer = NULL;
 491                         }
 492                         if (obj->type == OBJ_COMMIT) {
 493                                 struct commit *commit = (struct commit *) obj;
 494                                 commit->buffer = NULL;
 495                         }
 496                         obj->flags |= FLAG_CHECKED;
 497                 }
 498         }
 499 }
 500
 501 static int is_delta_type(enum object_type type)
 502 {
 503         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 504 }
 505
 506 static void *get_base_data(struct base_data *c)
 507 {
 508         if (!c->data) {
 509                 struct object_entry *obj = c->obj;
 510
 511                 if (is_delta_type(obj->type)) {
 512                         void *base = get_base_data(c->base);
 513                         void *raw = get_data_from_pack(obj);
 514                         c->data = patch_delta(
 515                                 base, c->base->size,
 516                                 raw, obj->size,
 517                                 &c->size);
 518                         free(raw);
 519                         if (!c->data)
 520                                 bad_object(obj->idx.offset, "failed to apply delta");
 521                 } else {
 522                         c->data = get_data_from_pack(obj);
 523                         c->size = obj->size;
 524                 }
 525
 526                 base_cache_used += c->size;
 527                 prune_base_data(c);
 528         }
 529         return c->data;
 530 }
 531
 532 static void resolve_delta(struct object_entry *delta_obj,
 533                           struct base_data *base, struct base_data *result)
 534 {
 535         void *base_data, *delta_data;
 536
 537         delta_obj->real_type = base->obj->real_type;
 538         delta_data = get_data_from_pack(delta_obj);
 539         base_data = get_base_data(base);
 540         result->obj = delta_obj;
 541         result->data = patch_delta(base_data, base->size,
 542                                    delta_data, delta_obj->size, &result->size);
 543         free(delta_data);
 544         if (!result->data)
 545                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 546         sha1_object(result->data, result->size, delta_obj->real_type,
 547                     delta_obj->idx.sha1);
 548         nr_resolved_deltas++;
 549 }
 550
 551 static void find_unresolved_deltas(struct base_data *base,
 552                                    struct base_data *prev_base)
 553 {
 554         int i, ref_first, ref_last, ofs_first, ofs_last;
 555
 556         /*
 557          * This is a recursive function. Those brackets should help reducing
 558          * stack usage by limiting the scope of the delta_base union.
 559          */
 560         {
 561                 union delta_base base_spec;
 562
 563                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 564                 find_delta_children(&base_spec,
 565                                     &ref_first, &ref_last, OBJ_REF_DELTA);
 566
 567                 memset(&base_spec, 0, sizeof(base_spec));
 568                 base_spec.offset = base->obj->idx.offset;
 569                 find_delta_children(&base_spec,
 570                                     &ofs_first, &ofs_last, OBJ_OFS_DELTA);
 571         }
 572
 573         if (ref_last == -1 && ofs_last == -1) {
 574                 free(base->data);
 575                 return;
 576         }
 577
 578         link_base_data(prev_base, base);
 579
 580         for (i = ref_first; i <= ref_last; i++) {
 581                 struct object_entry *child = objects + deltas[i].obj_no;
 582                 struct base_data result;
 583
 584                 assert(child->real_type == OBJ_REF_DELTA);
 585                 resolve_delta(child, base, &result);
 586                 if (i == ref_last && ofs_last == -1)
 587                         free_base_data(base);
 588                 find_unresolved_deltas(&result, base);
 589         }
 590
 591         for (i = ofs_first; i <= ofs_last; i++) {
 592                 struct object_entry *child = objects + deltas[i].obj_no;
 593                 struct base_data result;
 594
 595                 assert(child->real_type == OBJ_OFS_DELTA);
 596                 resolve_delta(child, base, &result);
 597                 if (i == ofs_last)
 598                         free_base_data(base);
 599                 find_unresolved_deltas(&result, base);
 600         }
 601
 602         unlink_base_data(base);
 603 }
 604
 605 static int compare_delta_entry(const void *a, const void *b)
 606 {
 607         const struct delta_entry *delta_a = a;
 608         const struct delta_entry *delta_b = b;
 609
 610         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 611         return compare_delta_bases(&delta_a->base, &delta_b->base,
 612                                    objects[delta_a->obj_no].type,
 613                                    objects[delta_b->obj_no].type);
 614 }
 615
 616 /* Parse all objects and return the pack content SHA1 hash */
 617 static void parse_pack_objects(unsigned char *sha1)
 618 {
 619         int i;
 620         struct delta_entry *delta = deltas;
 621         struct stat st;
 622
 623         /*
 624          * First pass:
 625          * - find locations of all objects;
 626          * - calculate SHA1 of all non-delta objects;
 627          * - remember base (SHA1 or offset) for all deltas.
 628          */
 629         if (verbose)
 630                 progress = start_progress(
 631                                 from_stdin ? "Receiving objects" : "Indexing objects",
 632                                 nr_objects);
 633         for (i = 0; i < nr_objects; i++) {
 634                 struct object_entry *obj = &objects[i];
 635                 void *data = unpack_raw_entry(obj, &delta->base);
 636                 obj->real_type = obj->type;
 637                 if (is_delta_type(obj->type)) {
 638                         nr_deltas++;
 639                         delta->obj_no = i;
 640                         delta++;
 641                 } else
 642                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 643                 free(data);
 644                 display_progress(progress, i+1);
 645         }
 646         objects[i].idx.offset = consumed_bytes;
 647         stop_progress(&progress);
 648
 649         /* Check pack integrity */
 650         flush();
 651         git_SHA1_Final(sha1, &input_ctx);
 652         if (hashcmp(fill(20), sha1))
 653                 die("pack is corrupted (SHA1 mismatch)");
 654         use(20);
 655
 656         /* If input_fd is a file, we should have reached its end now. */
 657         if (fstat(input_fd, &st))
 658                 die_errno("cannot fstat packfile");
 659         if (S_ISREG(st.st_mode) &&
 660                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 661                 die("pack has junk at the end");
 662
 663         if (!nr_deltas)
 664                 return;
 665
 666         /* Sort deltas by base SHA1/offset for fast searching */
 667         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 668               compare_delta_entry);
 669
 670         /*
 671          * Second pass:
 672          * - for all non-delta objects, look if it is used as a base for
 673          *   deltas;
 674          * - if used as a base, uncompress the object and apply all deltas,
 675          *   recursively checking if the resulting object is used as a base
 676          *   for some more deltas.
 677          */
 678         if (verbose)
 679                 progress = start_progress("Resolving deltas", nr_deltas);
 680         for (i = 0; i < nr_objects; i++) {
 681                 struct object_entry *obj = &objects[i];
 682                 struct base_data base_obj;
 683
 684                 if (is_delta_type(obj->type))
 685                         continue;
 686                 base_obj.obj = obj;
 687                 base_obj.data = NULL;
 688                 find_unresolved_deltas(&base_obj, NULL);
 689                 display_progress(progress, nr_resolved_deltas);
 690         }
 691 }
 692
 693 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 694 {
 695         z_stream stream;
 696         int status;
 697         unsigned char outbuf[4096];
 698
 699         memset(&stream, 0, sizeof(stream));
 700         deflateInit(&stream, zlib_compression_level);
 701         stream.next_in = in;
 702         stream.avail_in = size;
 703
 704         do {
 705                 stream.next_out = outbuf;
 706                 stream.avail_out = sizeof(outbuf);
 707                 status = deflate(&stream, Z_FINISH);
 708                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
 709         } while (status == Z_OK);
 710
 711         if (status != Z_STREAM_END)
 712                 die("unable to deflate appended object (%d)", status);
 713         size = stream.total_out;
 714         deflateEnd(&stream);
 715         return size;
 716 }
 717
 718 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 719                                const unsigned char *sha1, void *buf,
 720                                unsigned long size, enum object_type type)
 721 {
 722         struct object_entry *obj = &objects[nr_objects++];
 723         unsigned char header[10];
 724         unsigned long s = size;
 725         int n = 0;
 726         unsigned char c = (type << 4) | (s & 15);
 727         s >>= 4;
 728         while (s) {
 729                 header[n++] = c | 0x80;
 730                 c = s & 0x7f;
 731                 s >>= 7;
 732         }
 733         header[n++] = c;
 734         crc32_begin(f);
 735         sha1write(f, header, n);
 736         obj[0].size = size;
 737         obj[0].hdr_size = n;
 738         obj[0].type = type;
 739         obj[0].real_type = type;
 740         obj[1].idx.offset = obj[0].idx.offset + n;
 741         obj[1].idx.offset += write_compressed(f, buf, size);
 742         obj[0].idx.crc32 = crc32_end(f);
 743         sha1flush(f);
 744         hashcpy(obj->idx.sha1, sha1);
 745         return obj;
 746 }
 747
 748 static int delta_pos_compare(const void *_a, const void *_b)
 749 {
 750         struct delta_entry *a = *(struct delta_entry **)_a;
 751         struct delta_entry *b = *(struct delta_entry **)_b;
 752         return a->obj_no - b->obj_no;
 753 }
 754
 755 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 756 {
 757         struct delta_entry **sorted_by_pos;
 758         int i, n = 0;
 759
 760         /*
 761          * Since many unresolved deltas may well be themselves base objects
 762          * for more unresolved deltas, we really want to include the
 763          * smallest number of base objects that would cover as much delta
 764          * as possible by picking the
 765          * trunc deltas first, allowing for other deltas to resolve without
 766          * additional base objects.  Since most base objects are to be found
 767          * before deltas depending on them, a good heuristic is to start
 768          * resolving deltas in the same order as their position in the pack.
 769          */
 770         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 771         for (i = 0; i < nr_deltas; i++) {
 772                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 773                         continue;
 774                 sorted_by_pos[n++] = &deltas[i];
 775         }
 776         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 777
 778         for (i = 0; i < n; i++) {
 779                 struct delta_entry *d = sorted_by_pos[i];
 780                 enum object_type type;
 781                 struct base_data base_obj;
 782
 783                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 784                         continue;
 785                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 786                 if (!base_obj.data)
 787                         continue;
 788
 789                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 790                                 base_obj.size, typename(type)))
 791                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 792                 base_obj.obj = append_obj_to_pack(f, d->base.sha1,
 793                                         base_obj.data, base_obj.size, type);
 794                 find_unresolved_deltas(&base_obj, NULL);
 795                 display_progress(progress, nr_resolved_deltas);
 796         }
 797         free(sorted_by_pos);
 798 }
 799
 800 static void final(const char *final_pack_name, const char *curr_pack_name,
 801                   const char *final_index_name, const char *curr_index_name,
 802                   const char *keep_name, const char *keep_msg,
 803                   unsigned char *sha1)
 804 {
 805         const char *report = "pack";
 806         char name[PATH_MAX];
 807         int err;
 808
 809         if (!from_stdin) {
 810                 close(input_fd);
 811         } else {
 812                 fsync_or_die(output_fd, curr_pack_name);
 813                 err = close(output_fd);
 814                 if (err)
 815                         die_errno("error while closing pack file");
 816         }
 817
 818         if (keep_msg) {
 819                 int keep_fd, keep_msg_len = strlen(keep_msg);
 820
 821                 if (!keep_name)
 822                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
 823                 else
 824                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 825
 826                 if (keep_fd < 0) {
 827                         if (errno != EEXIST)
 828                                 die_errno("cannot write keep file '%s'",
 829                                           keep_name);
 830                 } else {
 831                         if (keep_msg_len > 0) {
 832                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 833                                 write_or_die(keep_fd, "\n", 1);
 834                         }
 835                         if (close(keep_fd) != 0)
 836                                 die_errno("cannot close written keep file '%s'",
 837                                     keep_name);
 838                         report = "keep";
 839                 }
 840         }
 841
 842         if (final_pack_name != curr_pack_name) {
 843                 if (!final_pack_name) {
 844                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 845                                  get_object_directory(), sha1_to_hex(sha1));
 846                         final_pack_name = name;
 847                 }
 848                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 849                         die("cannot store pack file");
 850         } else if (from_stdin)
 851                 chmod(final_pack_name, 0444);
 852
 853         if (final_index_name != curr_index_name) {
 854                 if (!final_index_name) {
 855                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 856                                  get_object_directory(), sha1_to_hex(sha1));
 857                         final_index_name = name;
 858                 }
 859                 if (move_temp_to_file(curr_index_name, final_index_name))
 860                         die("cannot store index file");
 861         } else
 862                 chmod(final_index_name, 0444);
 863
 864         if (!from_stdin) {
 865                 printf("%s\n", sha1_to_hex(sha1));
 866         } else {
 867                 char buf[48];
 868                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 869                                    report, sha1_to_hex(sha1));
 870                 write_or_die(1, buf, len);
 871
 872                 /*
 873                  * Let's just mimic git-unpack-objects here and write
 874                  * the last part of the input buffer to stdout.
 875                  */
 876                 while (input_len) {
 877                         err = xwrite(1, input_buffer + input_offset, input_len);
 878                         if (err <= 0)
 879                                 break;
 880                         input_len -= err;
 881                         input_offset += err;
 882                 }
 883         }
 884 }
 885
 886 static int git_index_pack_config(const char *k, const char *v, void *cb)
 887 {
 888         struct pack_idx_option *opts = cb;
 889
 890         if (!strcmp(k, "pack.indexversion")) {
 891                 opts->version = git_config_int(k, v);
 892                 if (opts->version > 2)
 893                         die("bad pack.indexversion=%"PRIu32, opts->version);
 894                 return 0;
 895         }
 896         return git_default_config(k, v, cb);
 897 }
 898
 899 static int cmp_uint32(const void *a_, const void *b_)
 900 {
 901         uint32_t a = *((uint32_t *)a_);
 902         uint32_t b = *((uint32_t *)b_);
 903
 904         return (a < b) ? -1 : (a != b);
 905 }
 906
 907 static void read_v2_anomalous_offsets(struct packed_git *p,
 908                                       struct pack_idx_option *opts)
 909 {
 910         const uint32_t *idx1, *idx2;
 911         uint32_t i;
 912
 913         /* The address of the 4-byte offset table */
 914         idx1 = (((const uint32_t *)p->index_data)
 915                 + 2 /* 8-byte header */
 916                 + 256 /* fan out */
 917                 + 5 * p->num_objects /* 20-byte SHA-1 table */
 918                 + p->num_objects /* CRC32 table */
 919                 );
 920
 921         /* The address of the 8-byte offset table */
 922         idx2 = idx1 + p->num_objects;
 923
 924         for (i = 0; i < p->num_objects; i++) {
 925                 uint32_t off = ntohl(idx1[i]);
 926                 if (!(off & 0x80000000))
 927                         continue;
 928                 off = off & 0x7fffffff;
 929                 if (idx2[off * 2])
 930                         continue;
 931                 /*
 932                  * The real offset is ntohl(idx2[off * 2]) in high 4
 933                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
 934                  * octets.  But idx2[off * 2] is Zero!!!
 935                  */
 936                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
 937                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
 938         }
 939
 940         if (1 < opts->anomaly_nr)
 941                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
 942 }
 943
 944 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
 945 {
 946         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
 947
 948         if (!p)
 949                 die("Cannot open existing pack file '%s'", pack_name);
 950         if (open_pack_index(p))
 951                 die("Cannot open existing pack idx file for '%s'", pack_name);
 952
 953         /* Read the attributes from the existing idx file */
 954         opts->version = p->index_version;
 955
 956         if (opts->version == 2)
 957                 read_v2_anomalous_offsets(p, opts);
 958
 959         /*
 960          * Get rid of the idx file as we do not need it anymore.
 961          * NEEDSWORK: extract this bit from free_pack_by_name() in
 962          * sha1_file.c, perhaps?  It shouldn't matter very much as we
 963          * know we haven't installed this pack (hence we never have
 964          * read anything from it).
 965          */
 966         close_pack_index(p);
 967         free(p);
 968 }
 969
 970 int cmd_index_pack(int argc, const char **argv, const char *prefix)
 971 {
 972         int i, fix_thin_pack = 0, verify = 0;
 973         const char *curr_pack, *curr_index;
 974         const char *index_name = NULL, *pack_name = NULL;
 975         const char *keep_name = NULL, *keep_msg = NULL;
 976         char *index_name_buf = NULL, *keep_name_buf = NULL;
 977         struct pack_idx_entry **idx_objects;
 978         struct pack_idx_option opts;
 979         unsigned char pack_sha1[20];
 980
 981         if (argc == 2 && !strcmp(argv[1], "-h"))
 982                 usage(index_pack_usage);
 983
 984         read_replace_refs = 0;
 985
 986         reset_pack_idx_option(&opts);
 987         git_config(git_index_pack_config, &opts);
 988         if (prefix && chdir(prefix))
 989                 die("Cannot come back to cwd");
 990
 991         for (i = 1; i < argc; i++) {
 992                 const char *arg = argv[i];
 993
 994                 if (*arg == '-') {
 995                         if (!strcmp(arg, "--stdin")) {
 996                                 from_stdin = 1;
 997                         } else if (!strcmp(arg, "--fix-thin")) {
 998                                 fix_thin_pack = 1;
 999                         } else if (!strcmp(arg, "--strict")) {
1000                                 strict = 1;
1001                         } else if (!strcmp(arg, "--verify")) {
1002                                 verify = 1;
1003                         } else if (!strcmp(arg, "--keep")) {
1004                                 keep_msg = "";
1005                         } else if (!prefixcmp(arg, "--keep=")) {
1006                                 keep_msg = arg + 7;
1007                         } else if (!prefixcmp(arg, "--pack_header=")) {
1008                                 struct pack_header *hdr;
1009                                 char *c;
1010
1011                                 hdr = (struct pack_header *)input_buffer;
1012                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1013                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1014                                 if (*c != ',')
1015                                         die("bad %s", arg);
1016                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1017                                 if (*c)
1018                                         die("bad %s", arg);
1019                                 input_len = sizeof(*hdr);
1020                         } else if (!strcmp(arg, "-v")) {
1021                                 verbose = 1;
1022                         } else if (!strcmp(arg, "-o")) {
1023                                 if (index_name || (i+1) >= argc)
1024                                         usage(index_pack_usage);
1025                                 index_name = argv[++i];
1026                         } else if (!prefixcmp(arg, "--index-version=")) {
1027                                 char *c;
1028                                 opts.version = strtoul(arg + 16, &c, 10);
1029                                 if (opts.version > 2)
1030                                         die("bad %s", arg);
1031                                 if (*c == ',')
1032                                         opts.off32_limit = strtoul(c+1, &c, 0);
1033                                 if (*c || opts.off32_limit & 0x80000000)
1034                                         die("bad %s", arg);
1035                         } else
1036                                 usage(index_pack_usage);
1037                         continue;
1038                 }
1039
1040                 if (pack_name)
1041                         usage(index_pack_usage);
1042                 pack_name = arg;
1043         }
1044
1045         if (!pack_name && !from_stdin)
1046                 usage(index_pack_usage);
1047         if (fix_thin_pack && !from_stdin)
1048                 die("--fix-thin cannot be used without --stdin");
1049         if (!index_name && pack_name) {
1050                 int len = strlen(pack_name);
1051                 if (!has_extension(pack_name, ".pack"))
1052                         die("packfile name '%s' does not end with '.pack'",
1053                             pack_name);
1054                 index_name_buf = xmalloc(len);
1055                 memcpy(index_name_buf, pack_name, len - 5);
1056                 strcpy(index_name_buf + len - 5, ".idx");
1057                 index_name = index_name_buf;
1058         }
1059         if (keep_msg && !keep_name && pack_name) {
1060                 int len = strlen(pack_name);
1061                 if (!has_extension(pack_name, ".pack"))
1062                         die("packfile name '%s' does not end with '.pack'",
1063                             pack_name);
1064                 keep_name_buf = xmalloc(len);
1065                 memcpy(keep_name_buf, pack_name, len - 5);
1066                 strcpy(keep_name_buf + len - 5, ".keep");
1067                 keep_name = keep_name_buf;
1068         }
1069         if (verify) {
1070                 if (!index_name)
1071                         die("--verify with no packfile name given");
1072                 read_idx_option(&opts, index_name);
1073                 opts.flags |= WRITE_IDX_VERIFY;
1074         }
1075
1076         curr_pack = open_pack_file(pack_name);
1077         parse_pack_header();
1078         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
1079         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
1080         parse_pack_objects(pack_sha1);
1081         if (nr_deltas == nr_resolved_deltas) {
1082                 stop_progress(&progress);
1083                 /* Flush remaining pack final 20-byte SHA1. */
1084                 flush();
1085         } else {
1086                 if (fix_thin_pack) {
1087                         struct sha1file *f;
1088                         unsigned char read_sha1[20], tail_sha1[20];
1089                         char msg[48];
1090                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
1091                         int nr_objects_initial = nr_objects;
1092                         if (nr_unresolved <= 0)
1093                                 die("confusion beyond insanity");
1094                         objects = xrealloc(objects,
1095                                            (nr_objects + nr_unresolved + 1)
1096                                            * sizeof(*objects));
1097                         f = sha1fd(output_fd, curr_pack);
1098                         fix_unresolved_deltas(f, nr_unresolved);
1099                         sprintf(msg, "completed with %d local objects",
1100                                 nr_objects - nr_objects_initial);
1101                         stop_progress_msg(&progress, msg);
1102                         sha1close(f, tail_sha1, 0);
1103                         hashcpy(read_sha1, pack_sha1);
1104                         fixup_pack_header_footer(output_fd, pack_sha1,
1105                                                  curr_pack, nr_objects,
1106                                                  read_sha1, consumed_bytes-20);
1107                         if (hashcmp(read_sha1, tail_sha1) != 0)
1108                                 die("Unexpected tail checksum for %s "
1109                                     "(disk corruption?)", curr_pack);
1110                 }
1111                 if (nr_deltas != nr_resolved_deltas)
1112                         die("pack has %d unresolved deltas",
1113                             nr_deltas - nr_resolved_deltas);
1114         }
1115         free(deltas);
1116         if (strict)
1117                 check_objects();
1118
1119         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1120         for (i = 0; i < nr_objects; i++)
1121                 idx_objects[i] = &objects[i].idx;
1122         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1123         free(idx_objects);
1124
1125         if (!verify)
1126                 final(pack_name, curr_pack,
1127                       index_name, curr_index,
1128                       keep_name, keep_msg,
1129                       pack_sha1);
1130         else
1131                 close(input_fd);
1132         free(objects);
1133         free(index_name_buf);
1134         free(keep_name_buf);
1135         if (pack_name == NULL)
1136                 free((void *) curr_pack);
1137         if (index_name == NULL)
1138                 free((void *) curr_index);
1139
1140         return 0;
1141 }