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
  13 static const char index_pack_usage[] =
  14 "git index-pack [-v] [-o <index-file>] [ --keep | --keep=<msg> ] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  15
  16 struct object_entry {
  17         struct pack_idx_entry idx;
  18         unsigned long size;
  19         unsigned int hdr_size;
  20         enum object_type type;
  21         enum object_type real_type;
  22 };
  23
  24 union delta_base {
  25         unsigned char sha1[20];
  26         off_t offset;
  27 };
  28
  29 struct base_data {
  30         struct base_data *base;
  31         struct base_data *child;
  32         struct object_entry *obj;
  33         void *data;
  34         unsigned long size;
  35 };
  36
  37 /*
  38  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  39  * to memcmp() only the first 20 bytes.
  40  */
  41 #define UNION_BASE_SZ   20
  42
  43 #define FLAG_LINK (1u<<20)
  44 #define FLAG_CHECKED (1u<<21)
  45
  46 struct delta_entry {
  47         union delta_base base;
  48         int obj_no;
  49 };
  50
  51 static struct object_entry *objects;
  52 static struct delta_entry *deltas;
  53 static struct base_data *base_cache;
  54 static size_t base_cache_used;
  55 static int nr_objects;
  56 static int nr_deltas;
  57 static int nr_resolved_deltas;
  58
  59 static int from_stdin;
  60 static int strict;
  61 static int verbose;
  62
  63 static struct progress *progress;
  64
  65 /* We always read in 4kB chunks. */
  66 static unsigned char input_buffer[4096];
  67 static unsigned int input_offset, input_len;
  68 static off_t consumed_bytes;
  69 static git_SHA_CTX input_ctx;
  70 static uint32_t input_crc32;
  71 static int input_fd, output_fd, pack_fd;
  72
  73 static int mark_link(struct object *obj, int type, void *data)
  74 {
  75         if (!obj)
  76                 return -1;
  77
  78         if (type != OBJ_ANY && obj->type != type)
  79                 die("object type mismatch at %s", sha1_to_hex(obj->sha1));
  80
  81         obj->flags |= FLAG_LINK;
  82         return 0;
  83 }
  84
  85 /* The content of each linked object must have been checked
  86    or it must be already present in the object database */
  87 static void check_object(struct object *obj)
  88 {
  89         if (!obj)
  90                 return;
  91
  92         if (!(obj->flags & FLAG_LINK))
  93                 return;
  94
  95         if (!(obj->flags & FLAG_CHECKED)) {
  96                 unsigned long size;
  97                 int type = sha1_object_info(obj->sha1, &size);
  98                 if (type != obj->type || type <= 0)
  99                         die("object of unexpected type");
 100                 obj->flags |= FLAG_CHECKED;
 101                 return;
 102         }
 103 }
 104
 105 static void check_objects(void)
 106 {
 107         unsigned i, max;
 108
 109         max = get_max_object_index();
 110         for (i = 0; i < max; i++)
 111                 check_object(get_indexed_object(i));
 112 }
 113
 114
 115 /* Discard current buffer used content. */
 116 static void flush(void)
 117 {
 118         if (input_offset) {
 119                 if (output_fd >= 0)
 120                         write_or_die(output_fd, input_buffer, input_offset);
 121                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 122                 memmove(input_buffer, input_buffer + input_offset, input_len);
 123                 input_offset = 0;
 124         }
 125 }
 126
 127 /*
 128  * Make sure at least "min" bytes are available in the buffer, and
 129  * return the pointer to the buffer.
 130  */
 131 static void *fill(int min)
 132 {
 133         if (min <= input_len)
 134                 return input_buffer + input_offset;
 135         if (min > sizeof(input_buffer))
 136                 die("cannot fill %d bytes", min);
 137         flush();
 138         do {
 139                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 140                                 sizeof(input_buffer) - input_len);
 141                 if (ret <= 0) {
 142                         if (!ret)
 143                                 die("early EOF");
 144                         die_errno("read error on input");
 145                 }
 146                 input_len += ret;
 147                 if (from_stdin)
 148                         display_throughput(progress, consumed_bytes + input_len);
 149         } while (input_len < min);
 150         return input_buffer;
 151 }
 152
 153 static void use(int bytes)
 154 {
 155         if (bytes > input_len)
 156                 die("used more bytes than were available");
 157         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 158         input_len -= bytes;
 159         input_offset += bytes;
 160
 161         /* make sure off_t is sufficiently large not to wrap */
 162         if (signed_add_overflows(consumed_bytes, bytes))
 163                 die("pack too large for current definition of off_t");
 164         consumed_bytes += bytes;
 165 }
 166
 167 static const char *open_pack_file(const char *pack_name)
 168 {
 169         if (from_stdin) {
 170                 input_fd = 0;
 171                 if (!pack_name) {
 172                         static char tmpfile[PATH_MAX];
 173                         output_fd = odb_mkstemp(tmpfile, sizeof(tmpfile),
 174                                                 "pack/tmp_pack_XXXXXX");
 175                         pack_name = xstrdup(tmpfile);
 176                 } else
 177                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 178                 if (output_fd < 0)
 179                         die_errno("unable to create '%s'", pack_name);
 180                 pack_fd = output_fd;
 181         } else {
 182                 input_fd = open(pack_name, O_RDONLY);
 183                 if (input_fd < 0)
 184                         die_errno("cannot open packfile '%s'", pack_name);
 185                 output_fd = -1;
 186                 pack_fd = input_fd;
 187         }
 188         git_SHA1_Init(&input_ctx);
 189         return pack_name;
 190 }
 191
 192 static void parse_pack_header(void)
 193 {
 194         struct pack_header *hdr = fill(sizeof(struct pack_header));
 195
 196         /* Header consistency check */
 197         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 198                 die("pack signature mismatch");
 199         if (!pack_version_ok(hdr->hdr_version))
 200                 die("pack version %"PRIu32" unsupported",
 201                         ntohl(hdr->hdr_version));
 202
 203         nr_objects = ntohl(hdr->hdr_entries);
 204         use(sizeof(struct pack_header));
 205 }
 206
 207 static NORETURN void bad_object(unsigned long offset, const char *format,
 208                        ...) __attribute__((format (printf, 2, 3)));
 209
 210 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 211 {
 212         va_list params;
 213         char buf[1024];
 214
 215         va_start(params, format);
 216         vsnprintf(buf, sizeof(buf), format, params);
 217         va_end(params);
 218         die("pack has bad object at offset %lu: %s", offset, buf);
 219 }
 220
 221 static void free_base_data(struct base_data *c)
 222 {
 223         if (c->data) {
 224                 free(c->data);
 225                 c->data = NULL;
 226                 base_cache_used -= c->size;
 227         }
 228 }
 229
 230 static void prune_base_data(struct base_data *retain)
 231 {
 232         struct base_data *b;
 233         for (b = base_cache;
 234              base_cache_used > delta_base_cache_limit && b;
 235              b = b->child) {
 236                 if (b->data && b != retain)
 237                         free_base_data(b);
 238         }
 239 }
 240
 241 static void link_base_data(struct base_data *base, struct base_data *c)
 242 {
 243         if (base)
 244                 base->child = c;
 245         else
 246                 base_cache = c;
 247
 248         c->base = base;
 249         c->child = NULL;
 250         if (c->data)
 251                 base_cache_used += c->size;
 252         prune_base_data(c);
 253 }
 254
 255 static void unlink_base_data(struct base_data *c)
 256 {
 257         struct base_data *base = c->base;
 258         if (base)
 259                 base->child = NULL;
 260         else
 261                 base_cache = NULL;
 262         free_base_data(c);
 263 }
 264
 265 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 266 {
 267         int status;
 268         z_stream stream;
 269         void *buf = xmalloc(size);
 270
 271         memset(&stream, 0, sizeof(stream));
 272         git_inflate_init(&stream);
 273         stream.next_out = buf;
 274         stream.avail_out = size;
 275
 276         do {
 277                 stream.next_in = fill(1);
 278                 stream.avail_in = input_len;
 279                 status = git_inflate(&stream, 0);
 280                 use(input_len - stream.avail_in);
 281         } while (status == Z_OK);
 282         if (stream.total_out != size || status != Z_STREAM_END)
 283                 bad_object(offset, "inflate returned %d", status);
 284         git_inflate_end(&stream);
 285         return buf;
 286 }
 287
 288 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 289 {
 290         unsigned char *p;
 291         unsigned long size, c;
 292         off_t base_offset;
 293         unsigned shift;
 294         void *data;
 295
 296         obj->idx.offset = consumed_bytes;
 297         input_crc32 = crc32(0, NULL, 0);
 298
 299         p = fill(1);
 300         c = *p;
 301         use(1);
 302         obj->type = (c >> 4) & 7;
 303         size = (c & 15);
 304         shift = 4;
 305         while (c & 0x80) {
 306                 p = fill(1);
 307                 c = *p;
 308                 use(1);
 309                 size += (c & 0x7f) << shift;
 310                 shift += 7;
 311         }
 312         obj->size = size;
 313
 314         switch (obj->type) {
 315         case OBJ_REF_DELTA:
 316                 hashcpy(delta_base->sha1, fill(20));
 317                 use(20);
 318                 break;
 319         case OBJ_OFS_DELTA:
 320                 memset(delta_base, 0, sizeof(*delta_base));
 321                 p = fill(1);
 322                 c = *p;
 323                 use(1);
 324                 base_offset = c & 127;
 325                 while (c & 128) {
 326                         base_offset += 1;
 327                         if (!base_offset || MSB(base_offset, 7))
 328                                 bad_object(obj->idx.offset, "offset value overflow for delta base object");
 329                         p = fill(1);
 330                         c = *p;
 331                         use(1);
 332                         base_offset = (base_offset << 7) + (c & 127);
 333                 }
 334                 delta_base->offset = obj->idx.offset - base_offset;
 335                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 336                         bad_object(obj->idx.offset, "delta base offset is out of bound");
 337                 break;
 338         case OBJ_COMMIT:
 339         case OBJ_TREE:
 340         case OBJ_BLOB:
 341         case OBJ_TAG:
 342                 break;
 343         default:
 344                 bad_object(obj->idx.offset, "unknown object type %d", obj->type);
 345         }
 346         obj->hdr_size = consumed_bytes - obj->idx.offset;
 347
 348         data = unpack_entry_data(obj->idx.offset, obj->size);
 349         obj->idx.crc32 = input_crc32;
 350         return data;
 351 }
 352
 353 static void *get_data_from_pack(struct object_entry *obj)
 354 {
 355         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 356         unsigned long len = obj[1].idx.offset - from;
 357         unsigned char *data, *inbuf;
 358         z_stream stream;
 359         int status;
 360
 361         data = xmalloc(obj->size);
 362         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 363
 364         memset(&stream, 0, sizeof(stream));
 365         git_inflate_init(&stream);
 366         stream.next_out = data;
 367         stream.avail_out = obj->size;
 368
 369         do {
 370                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 371                 n = pread(pack_fd, inbuf, n, from);
 372                 if (n < 0)
 373                         die_errno("cannot pread pack file");
 374                 if (!n)
 375                         die("premature end of pack file, %lu bytes missing", len);
 376                 from += n;
 377                 len -= n;
 378                 stream.next_in = inbuf;
 379                 stream.avail_in = n;
 380                 status = git_inflate(&stream, 0);
 381         } while (len && status == Z_OK && !stream.avail_in);
 382
 383         /* This has been inflated OK when first encountered, so... */
 384         if (status != Z_STREAM_END || stream.total_out != obj->size)
 385                 die("serious inflate inconsistency");
 386
 387         git_inflate_end(&stream);
 388         free(inbuf);
 389         return data;
 390 }
 391
 392 static int find_delta(const union delta_base *base)
 393 {
 394         int first = 0, last = nr_deltas;
 395
 396         while (first < last) {
 397                 int next = (first + last) / 2;
 398                 struct delta_entry *delta = &deltas[next];
 399                 int cmp;
 400
 401                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 402                 if (!cmp)
 403                         return next;
 404                 if (cmp < 0) {
 405                         last = next;
 406                         continue;
 407                 }
 408                 first = next+1;
 409         }
 410         return -first-1;
 411 }
 412
 413 static void find_delta_children(const union delta_base *base,
 414                                 int *first_index, int *last_index)
 415 {
 416         int first = find_delta(base);
 417         int last = first;
 418         int end = nr_deltas - 1;
 419
 420         if (first < 0) {
 421                 *first_index = 0;
 422                 *last_index = -1;
 423                 return;
 424         }
 425         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 426                 --first;
 427         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 428                 ++last;
 429         *first_index = first;
 430         *last_index = last;
 431 }
 432
 433 static void sha1_object(const void *data, unsigned long size,
 434                         enum object_type type, unsigned char *sha1)
 435 {
 436         hash_sha1_file(data, size, typename(type), sha1);
 437         if (has_sha1_file(sha1)) {
 438                 void *has_data;
 439                 enum object_type has_type;
 440                 unsigned long has_size;
 441                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 442                 if (!has_data)
 443                         die("cannot read existing object %s", sha1_to_hex(sha1));
 444                 if (size != has_size || type != has_type ||
 445                     memcmp(data, has_data, size) != 0)
 446                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 447                 free(has_data);
 448         }
 449         if (strict) {
 450                 if (type == OBJ_BLOB) {
 451                         struct blob *blob = lookup_blob(sha1);
 452                         if (blob)
 453                                 blob->object.flags |= FLAG_CHECKED;
 454                         else
 455                                 die("invalid blob object %s", sha1_to_hex(sha1));
 456                 } else {
 457                         struct object *obj;
 458                         int eaten;
 459                         void *buf = (void *) data;
 460
 461                         /*
 462                          * we do not need to free the memory here, as the
 463                          * buf is deleted by the caller.
 464                          */
 465                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 466                         if (!obj)
 467                                 die("invalid %s", typename(type));
 468                         if (fsck_object(obj, 1, fsck_error_function))
 469                                 die("Error in object");
 470                         if (fsck_walk(obj, mark_link, NULL))
 471                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 472
 473                         if (obj->type == OBJ_TREE) {
 474                                 struct tree *item = (struct tree *) obj;
 475                                 item->buffer = NULL;
 476                         }
 477                         if (obj->type == OBJ_COMMIT) {
 478                                 struct commit *commit = (struct commit *) obj;
 479                                 commit->buffer = NULL;
 480                         }
 481                         obj->flags |= FLAG_CHECKED;
 482                 }
 483         }
 484 }
 485
 486 static void *get_base_data(struct base_data *c)
 487 {
 488         if (!c->data) {
 489                 struct object_entry *obj = c->obj;
 490
 491                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 492                         void *base = get_base_data(c->base);
 493                         void *raw = get_data_from_pack(obj);
 494                         c->data = patch_delta(
 495                                 base, c->base->size,
 496                                 raw, obj->size,
 497                                 &c->size);
 498                         free(raw);
 499                         if (!c->data)
 500                                 bad_object(obj->idx.offset, "failed to apply delta");
 501                 } else {
 502                         c->data = get_data_from_pack(obj);
 503                         c->size = obj->size;
 504                 }
 505
 506                 base_cache_used += c->size;
 507                 prune_base_data(c);
 508         }
 509         return c->data;
 510 }
 511
 512 static void resolve_delta(struct object_entry *delta_obj,
 513                           struct base_data *base, struct base_data *result)
 514 {
 515         void *base_data, *delta_data;
 516
 517         delta_obj->real_type = base->obj->real_type;
 518         delta_data = get_data_from_pack(delta_obj);
 519         base_data = get_base_data(base);
 520         result->obj = delta_obj;
 521         result->data = patch_delta(base_data, base->size,
 522                                    delta_data, delta_obj->size, &result->size);
 523         free(delta_data);
 524         if (!result->data)
 525                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 526         sha1_object(result->data, result->size, delta_obj->real_type,
 527                     delta_obj->idx.sha1);
 528         nr_resolved_deltas++;
 529 }
 530
 531 static void find_unresolved_deltas(struct base_data *base,
 532                                    struct base_data *prev_base)
 533 {
 534         int i, ref_first, ref_last, ofs_first, ofs_last;
 535
 536         /*
 537          * This is a recursive function. Those brackets should help reducing
 538          * stack usage by limiting the scope of the delta_base union.
 539          */
 540         {
 541                 union delta_base base_spec;
 542
 543                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 544                 find_delta_children(&base_spec, &ref_first, &ref_last);
 545
 546                 memset(&base_spec, 0, sizeof(base_spec));
 547                 base_spec.offset = base->obj->idx.offset;
 548                 find_delta_children(&base_spec, &ofs_first, &ofs_last);
 549         }
 550
 551         if (ref_last == -1 && ofs_last == -1) {
 552                 free(base->data);
 553                 return;
 554         }
 555
 556         link_base_data(prev_base, base);
 557
 558         for (i = ref_first; i <= ref_last; i++) {
 559                 struct object_entry *child = objects + deltas[i].obj_no;
 560                 if (child->real_type == OBJ_REF_DELTA) {
 561                         struct base_data result;
 562                         resolve_delta(child, base, &result);
 563                         if (i == ref_last && ofs_last == -1)
 564                                 free_base_data(base);
 565                         find_unresolved_deltas(&result, base);
 566                 }
 567         }
 568
 569         for (i = ofs_first; i <= ofs_last; i++) {
 570                 struct object_entry *child = objects + deltas[i].obj_no;
 571                 if (child->real_type == OBJ_OFS_DELTA) {
 572                         struct base_data result;
 573                         resolve_delta(child, base, &result);
 574                         if (i == ofs_last)
 575                                 free_base_data(base);
 576                         find_unresolved_deltas(&result, base);
 577                 }
 578         }
 579
 580         unlink_base_data(base);
 581 }
 582
 583 static int compare_delta_entry(const void *a, const void *b)
 584 {
 585         const struct delta_entry *delta_a = a;
 586         const struct delta_entry *delta_b = b;
 587         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 588 }
 589
 590 /* Parse all objects and return the pack content SHA1 hash */
 591 static void parse_pack_objects(unsigned char *sha1)
 592 {
 593         int i;
 594         struct delta_entry *delta = deltas;
 595         struct stat st;
 596
 597         /*
 598          * First pass:
 599          * - find locations of all objects;
 600          * - calculate SHA1 of all non-delta objects;
 601          * - remember base (SHA1 or offset) for all deltas.
 602          */
 603         if (verbose)
 604                 progress = start_progress(
 605                                 from_stdin ? "Receiving objects" : "Indexing objects",
 606                                 nr_objects);
 607         for (i = 0; i < nr_objects; i++) {
 608                 struct object_entry *obj = &objects[i];
 609                 void *data = unpack_raw_entry(obj, &delta->base);
 610                 obj->real_type = obj->type;
 611                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 612                         nr_deltas++;
 613                         delta->obj_no = i;
 614                         delta++;
 615                 } else
 616                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 617                 free(data);
 618                 display_progress(progress, i+1);
 619         }
 620         objects[i].idx.offset = consumed_bytes;
 621         stop_progress(&progress);
 622
 623         /* Check pack integrity */
 624         flush();
 625         git_SHA1_Final(sha1, &input_ctx);
 626         if (hashcmp(fill(20), sha1))
 627                 die("pack is corrupted (SHA1 mismatch)");
 628         use(20);
 629
 630         /* If input_fd is a file, we should have reached its end now. */
 631         if (fstat(input_fd, &st))
 632                 die_errno("cannot fstat packfile");
 633         if (S_ISREG(st.st_mode) &&
 634                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 635                 die("pack has junk at the end");
 636
 637         if (!nr_deltas)
 638                 return;
 639
 640         /* Sort deltas by base SHA1/offset for fast searching */
 641         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 642               compare_delta_entry);
 643
 644         /*
 645          * Second pass:
 646          * - for all non-delta objects, look if it is used as a base for
 647          *   deltas;
 648          * - if used as a base, uncompress the object and apply all deltas,
 649          *   recursively checking if the resulting object is used as a base
 650          *   for some more deltas.
 651          */
 652         if (verbose)
 653                 progress = start_progress("Resolving deltas", nr_deltas);
 654         for (i = 0; i < nr_objects; i++) {
 655                 struct object_entry *obj = &objects[i];
 656                 struct base_data base_obj;
 657
 658                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 659                         continue;
 660                 base_obj.obj = obj;
 661                 base_obj.data = NULL;
 662                 find_unresolved_deltas(&base_obj, NULL);
 663                 display_progress(progress, nr_resolved_deltas);
 664         }
 665 }
 666
 667 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 668 {
 669         z_stream stream;
 670         int status;
 671         unsigned char outbuf[4096];
 672
 673         memset(&stream, 0, sizeof(stream));
 674         deflateInit(&stream, zlib_compression_level);
 675         stream.next_in = in;
 676         stream.avail_in = size;
 677
 678         do {
 679                 stream.next_out = outbuf;
 680                 stream.avail_out = sizeof(outbuf);
 681                 status = deflate(&stream, Z_FINISH);
 682                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
 683         } while (status == Z_OK);
 684
 685         if (status != Z_STREAM_END)
 686                 die("unable to deflate appended object (%d)", status);
 687         size = stream.total_out;
 688         deflateEnd(&stream);
 689         return size;
 690 }
 691
 692 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 693                                const unsigned char *sha1, void *buf,
 694                                unsigned long size, enum object_type type)
 695 {
 696         struct object_entry *obj = &objects[nr_objects++];
 697         unsigned char header[10];
 698         unsigned long s = size;
 699         int n = 0;
 700         unsigned char c = (type << 4) | (s & 15);
 701         s >>= 4;
 702         while (s) {
 703                 header[n++] = c | 0x80;
 704                 c = s & 0x7f;
 705                 s >>= 7;
 706         }
 707         header[n++] = c;
 708         crc32_begin(f);
 709         sha1write(f, header, n);
 710         obj[0].size = size;
 711         obj[0].hdr_size = n;
 712         obj[0].type = type;
 713         obj[0].real_type = type;
 714         obj[1].idx.offset = obj[0].idx.offset + n;
 715         obj[1].idx.offset += write_compressed(f, buf, size);
 716         obj[0].idx.crc32 = crc32_end(f);
 717         sha1flush(f);
 718         hashcpy(obj->idx.sha1, sha1);
 719         return obj;
 720 }
 721
 722 static int delta_pos_compare(const void *_a, const void *_b)
 723 {
 724         struct delta_entry *a = *(struct delta_entry **)_a;
 725         struct delta_entry *b = *(struct delta_entry **)_b;
 726         return a->obj_no - b->obj_no;
 727 }
 728
 729 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 730 {
 731         struct delta_entry **sorted_by_pos;
 732         int i, n = 0;
 733
 734         /*
 735          * Since many unresolved deltas may well be themselves base objects
 736          * for more unresolved deltas, we really want to include the
 737          * smallest number of base objects that would cover as much delta
 738          * as possible by picking the
 739          * trunc deltas first, allowing for other deltas to resolve without
 740          * additional base objects.  Since most base objects are to be found
 741          * before deltas depending on them, a good heuristic is to start
 742          * resolving deltas in the same order as their position in the pack.
 743          */
 744         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 745         for (i = 0; i < nr_deltas; i++) {
 746                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 747                         continue;
 748                 sorted_by_pos[n++] = &deltas[i];
 749         }
 750         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 751
 752         for (i = 0; i < n; i++) {
 753                 struct delta_entry *d = sorted_by_pos[i];
 754                 enum object_type type;
 755                 struct base_data base_obj;
 756
 757                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 758                         continue;
 759                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 760                 if (!base_obj.data)
 761                         continue;
 762
 763                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 764                                 base_obj.size, typename(type)))
 765                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 766                 base_obj.obj = append_obj_to_pack(f, d->base.sha1,
 767                                         base_obj.data, base_obj.size, type);
 768                 find_unresolved_deltas(&base_obj, NULL);
 769                 display_progress(progress, nr_resolved_deltas);
 770         }
 771         free(sorted_by_pos);
 772 }
 773
 774 static void final(const char *final_pack_name, const char *curr_pack_name,
 775                   const char *final_index_name, const char *curr_index_name,
 776                   const char *keep_name, const char *keep_msg,
 777                   unsigned char *sha1)
 778 {
 779         const char *report = "pack";
 780         char name[PATH_MAX];
 781         int err;
 782
 783         if (!from_stdin) {
 784                 close(input_fd);
 785         } else {
 786                 fsync_or_die(output_fd, curr_pack_name);
 787                 err = close(output_fd);
 788                 if (err)
 789                         die_errno("error while closing pack file");
 790         }
 791
 792         if (keep_msg) {
 793                 int keep_fd, keep_msg_len = strlen(keep_msg);
 794
 795                 if (!keep_name)
 796                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
 797                 else
 798                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 799
 800                 if (keep_fd < 0) {
 801                         if (errno != EEXIST)
 802                                 die_errno("cannot write keep file '%s'",
 803                                           keep_name);
 804                 } else {
 805                         if (keep_msg_len > 0) {
 806                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 807                                 write_or_die(keep_fd, "\n", 1);
 808                         }
 809                         if (close(keep_fd) != 0)
 810                                 die_errno("cannot close written keep file '%s'",
 811                                     keep_name);
 812                         report = "keep";
 813                 }
 814         }
 815
 816         if (final_pack_name != curr_pack_name) {
 817                 if (!final_pack_name) {
 818                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 819                                  get_object_directory(), sha1_to_hex(sha1));
 820                         final_pack_name = name;
 821                 }
 822                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 823                         die("cannot store pack file");
 824         } else if (from_stdin)
 825                 chmod(final_pack_name, 0444);
 826
 827         if (final_index_name != curr_index_name) {
 828                 if (!final_index_name) {
 829                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 830                                  get_object_directory(), sha1_to_hex(sha1));
 831                         final_index_name = name;
 832                 }
 833                 if (move_temp_to_file(curr_index_name, final_index_name))
 834                         die("cannot store index file");
 835         } else
 836                 chmod(final_index_name, 0444);
 837
 838         if (!from_stdin) {
 839                 printf("%s\n", sha1_to_hex(sha1));
 840         } else {
 841                 char buf[48];
 842                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 843                                    report, sha1_to_hex(sha1));
 844                 write_or_die(1, buf, len);
 845
 846                 /*
 847                  * Let's just mimic git-unpack-objects here and write
 848                  * the last part of the input buffer to stdout.
 849                  */
 850                 while (input_len) {
 851                         err = xwrite(1, input_buffer + input_offset, input_len);
 852                         if (err <= 0)
 853                                 break;
 854                         input_len -= err;
 855                         input_offset += err;
 856                 }
 857         }
 858 }
 859
 860 static int git_index_pack_config(const char *k, const char *v, void *cb)
 861 {
 862         if (!strcmp(k, "pack.indexversion")) {
 863                 pack_idx_default_version = git_config_int(k, v);
 864                 if (pack_idx_default_version > 2)
 865                         die("bad pack.indexversion=%"PRIu32,
 866                                 pack_idx_default_version);
 867                 return 0;
 868         }
 869         return git_default_config(k, v, cb);
 870 }
 871
 872 int cmd_index_pack(int argc, const char **argv, const char *prefix)
 873 {
 874         int i, fix_thin_pack = 0;
 875         const char *curr_pack, *curr_index;
 876         const char *index_name = NULL, *pack_name = NULL;
 877         const char *keep_name = NULL, *keep_msg = NULL;
 878         char *index_name_buf = NULL, *keep_name_buf = NULL;
 879         struct pack_idx_entry **idx_objects;
 880         unsigned char pack_sha1[20];
 881
 882         if (argc == 2 && !strcmp(argv[1], "-h"))
 883                 usage(index_pack_usage);
 884
 885         read_replace_refs = 0;
 886
 887         git_config(git_index_pack_config, NULL);
 888         if (prefix && chdir(prefix))
 889                 die("Cannot come back to cwd");
 890
 891         for (i = 1; i < argc; i++) {
 892                 const char *arg = argv[i];
 893
 894                 if (*arg == '-') {
 895                         if (!strcmp(arg, "--stdin")) {
 896                                 from_stdin = 1;
 897                         } else if (!strcmp(arg, "--fix-thin")) {
 898                                 fix_thin_pack = 1;
 899                         } else if (!strcmp(arg, "--strict")) {
 900                                 strict = 1;
 901                         } else if (!strcmp(arg, "--keep")) {
 902                                 keep_msg = "";
 903                         } else if (!prefixcmp(arg, "--keep=")) {
 904                                 keep_msg = arg + 7;
 905                         } else if (!prefixcmp(arg, "--pack_header=")) {
 906                                 struct pack_header *hdr;
 907                                 char *c;
 908
 909                                 hdr = (struct pack_header *)input_buffer;
 910                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 911                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 912                                 if (*c != ',')
 913                                         die("bad %s", arg);
 914                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 915                                 if (*c)
 916                                         die("bad %s", arg);
 917                                 input_len = sizeof(*hdr);
 918                         } else if (!strcmp(arg, "-v")) {
 919                                 verbose = 1;
 920                         } else if (!strcmp(arg, "-o")) {
 921                                 if (index_name || (i+1) >= argc)
 922                                         usage(index_pack_usage);
 923                                 index_name = argv[++i];
 924                         } else if (!prefixcmp(arg, "--index-version=")) {
 925                                 char *c;
 926                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 927                                 if (pack_idx_default_version > 2)
 928                                         die("bad %s", arg);
 929                                 if (*c == ',')
 930                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 931                                 if (*c || pack_idx_off32_limit & 0x80000000)
 932                                         die("bad %s", arg);
 933                         } else
 934                                 usage(index_pack_usage);
 935                         continue;
 936                 }
 937
 938                 if (pack_name)
 939                         usage(index_pack_usage);
 940                 pack_name = arg;
 941         }
 942
 943         if (!pack_name && !from_stdin)
 944                 usage(index_pack_usage);
 945         if (fix_thin_pack && !from_stdin)
 946                 die("--fix-thin cannot be used without --stdin");
 947         if (!index_name && pack_name) {
 948                 int len = strlen(pack_name);
 949                 if (!has_extension(pack_name, ".pack"))
 950                         die("packfile name '%s' does not end with '.pack'",
 951                             pack_name);
 952                 index_name_buf = xmalloc(len);
 953                 memcpy(index_name_buf, pack_name, len - 5);
 954                 strcpy(index_name_buf + len - 5, ".idx");
 955                 index_name = index_name_buf;
 956         }
 957         if (keep_msg && !keep_name && pack_name) {
 958                 int len = strlen(pack_name);
 959                 if (!has_extension(pack_name, ".pack"))
 960                         die("packfile name '%s' does not end with '.pack'",
 961                             pack_name);
 962                 keep_name_buf = xmalloc(len);
 963                 memcpy(keep_name_buf, pack_name, len - 5);
 964                 strcpy(keep_name_buf + len - 5, ".keep");
 965                 keep_name = keep_name_buf;
 966         }
 967
 968         curr_pack = open_pack_file(pack_name);
 969         parse_pack_header();
 970         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 971         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 972         parse_pack_objects(pack_sha1);
 973         if (nr_deltas == nr_resolved_deltas) {
 974                 stop_progress(&progress);
 975                 /* Flush remaining pack final 20-byte SHA1. */
 976                 flush();
 977         } else {
 978                 if (fix_thin_pack) {
 979                         struct sha1file *f;
 980                         unsigned char read_sha1[20], tail_sha1[20];
 981                         char msg[48];
 982                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 983                         int nr_objects_initial = nr_objects;
 984                         if (nr_unresolved <= 0)
 985                                 die("confusion beyond insanity");
 986                         objects = xrealloc(objects,
 987                                            (nr_objects + nr_unresolved + 1)
 988                                            * sizeof(*objects));
 989                         f = sha1fd(output_fd, curr_pack);
 990                         fix_unresolved_deltas(f, nr_unresolved);
 991                         sprintf(msg, "completed with %d local objects",
 992                                 nr_objects - nr_objects_initial);
 993                         stop_progress_msg(&progress, msg);
 994                         sha1close(f, tail_sha1, 0);
 995                         hashcpy(read_sha1, pack_sha1);
 996                         fixup_pack_header_footer(output_fd, pack_sha1,
 997                                                  curr_pack, nr_objects,
 998                                                  read_sha1, consumed_bytes-20);
 999                         if (hashcmp(read_sha1, tail_sha1) != 0)
1000                                 die("Unexpected tail checksum for %s "
1001                                     "(disk corruption?)", curr_pack);
1002                 }
1003                 if (nr_deltas != nr_resolved_deltas)
1004                         die("pack has %d unresolved deltas",
1005                             nr_deltas - nr_resolved_deltas);
1006         }
1007         free(deltas);
1008         if (strict)
1009                 check_objects();
1010
1011         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1012         for (i = 0; i < nr_objects; i++)
1013                 idx_objects[i] = &objects[i].idx;
1014         curr_index = write_idx_file(index_name, idx_objects, nr_objects, pack_sha1);
1015         free(idx_objects);
1016
1017         final(pack_name, curr_pack,
1018                 index_name, curr_index,
1019                 keep_name, keep_msg,
1020                 pack_sha1);
1021         free(objects);
1022         free(index_name_buf);
1023         free(keep_name_buf);
1024         if (pack_name == NULL)
1025                 free((void *) curr_pack);
1026         if (index_name == NULL)
1027                 free((void *) curr_index);
1028
1029         return 0;
1030 }