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> }] [--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 (consumed_bytes > 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 find_delta(const union delta_base *base)
 395 {
 396         int first = 0, last = nr_deltas;
 397
 398         while (first < last) {
 399                 int next = (first + last) / 2;
 400                 struct delta_entry *delta = &deltas[next];
 401                 int cmp;
 402
 403                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 404                 if (!cmp)
 405                         return next;
 406                 if (cmp < 0) {
 407                         last = next;
 408                         continue;
 409                 }
 410                 first = next+1;
 411         }
 412         return -first-1;
 413 }
 414
 415 static void find_delta_children(const union delta_base *base,
 416                                 int *first_index, int *last_index)
 417 {
 418         int first = find_delta(base);
 419         int last = first;
 420         int end = nr_deltas - 1;
 421
 422         if (first < 0) {
 423                 *first_index = 0;
 424                 *last_index = -1;
 425                 return;
 426         }
 427         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 428                 --first;
 429         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 430                 ++last;
 431         *first_index = first;
 432         *last_index = last;
 433 }
 434
 435 static void sha1_object(const void *data, unsigned long size,
 436                         enum object_type type, unsigned char *sha1)
 437 {
 438         hash_sha1_file(data, size, typename(type), sha1);
 439         if (has_sha1_file(sha1)) {
 440                 void *has_data;
 441                 enum object_type has_type;
 442                 unsigned long has_size;
 443                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 444                 if (!has_data)
 445                         die("cannot read existing object %s", sha1_to_hex(sha1));
 446                 if (size != has_size || type != has_type ||
 447                     memcmp(data, has_data, size) != 0)
 448                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 449                 free(has_data);
 450         }
 451         if (strict) {
 452                 if (type == OBJ_BLOB) {
 453                         struct blob *blob = lookup_blob(sha1);
 454                         if (blob)
 455                                 blob->object.flags |= FLAG_CHECKED;
 456                         else
 457                                 die("invalid blob object %s", sha1_to_hex(sha1));
 458                 } else {
 459                         struct object *obj;
 460                         int eaten;
 461                         void *buf = (void *) data;
 462
 463                         /*
 464                          * we do not need to free the memory here, as the
 465                          * buf is deleted by the caller.
 466                          */
 467                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 468                         if (!obj)
 469                                 die("invalid %s", typename(type));
 470                         if (fsck_object(obj, 1, fsck_error_function))
 471                                 die("Error in object");
 472                         if (fsck_walk(obj, mark_link, NULL))
 473                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 474
 475                         if (obj->type == OBJ_TREE) {
 476                                 struct tree *item = (struct tree *) obj;
 477                                 item->buffer = NULL;
 478                         }
 479                         if (obj->type == OBJ_COMMIT) {
 480                                 struct commit *commit = (struct commit *) obj;
 481                                 commit->buffer = NULL;
 482                         }
 483                         obj->flags |= FLAG_CHECKED;
 484                 }
 485         }
 486 }
 487
 488 static void *get_base_data(struct base_data *c)
 489 {
 490         if (!c->data) {
 491                 struct object_entry *obj = c->obj;
 492
 493                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 494                         void *base = get_base_data(c->base);
 495                         void *raw = get_data_from_pack(obj);
 496                         c->data = patch_delta(
 497                                 base, c->base->size,
 498                                 raw, obj->size,
 499                                 &c->size);
 500                         free(raw);
 501                         if (!c->data)
 502                                 bad_object(obj->idx.offset, "failed to apply delta");
 503                 } else {
 504                         c->data = get_data_from_pack(obj);
 505                         c->size = obj->size;
 506                 }
 507
 508                 base_cache_used += c->size;
 509                 prune_base_data(c);
 510         }
 511         return c->data;
 512 }
 513
 514 static void resolve_delta(struct object_entry *delta_obj,
 515                           struct base_data *base, struct base_data *result)
 516 {
 517         void *base_data, *delta_data;
 518
 519         delta_obj->real_type = base->obj->real_type;
 520         delta_data = get_data_from_pack(delta_obj);
 521         base_data = get_base_data(base);
 522         result->obj = delta_obj;
 523         result->data = patch_delta(base_data, base->size,
 524                                    delta_data, delta_obj->size, &result->size);
 525         free(delta_data);
 526         if (!result->data)
 527                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 528         sha1_object(result->data, result->size, delta_obj->real_type,
 529                     delta_obj->idx.sha1);
 530         nr_resolved_deltas++;
 531 }
 532
 533 static void find_unresolved_deltas(struct base_data *base,
 534                                    struct base_data *prev_base)
 535 {
 536         int i, ref_first, ref_last, ofs_first, ofs_last;
 537
 538         /*
 539          * This is a recursive function. Those brackets should help reducing
 540          * stack usage by limiting the scope of the delta_base union.
 541          */
 542         {
 543                 union delta_base base_spec;
 544
 545                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 546                 find_delta_children(&base_spec, &ref_first, &ref_last);
 547
 548                 memset(&base_spec, 0, sizeof(base_spec));
 549                 base_spec.offset = base->obj->idx.offset;
 550                 find_delta_children(&base_spec, &ofs_first, &ofs_last);
 551         }
 552
 553         if (ref_last == -1 && ofs_last == -1) {
 554                 free(base->data);
 555                 return;
 556         }
 557
 558         link_base_data(prev_base, base);
 559
 560         for (i = ref_first; i <= ref_last; i++) {
 561                 struct object_entry *child = objects + deltas[i].obj_no;
 562                 if (child->real_type == OBJ_REF_DELTA) {
 563                         struct base_data result;
 564                         resolve_delta(child, base, &result);
 565                         if (i == ref_last && ofs_last == -1)
 566                                 free_base_data(base);
 567                         find_unresolved_deltas(&result, base);
 568                 }
 569         }
 570
 571         for (i = ofs_first; i <= ofs_last; i++) {
 572                 struct object_entry *child = objects + deltas[i].obj_no;
 573                 if (child->real_type == OBJ_OFS_DELTA) {
 574                         struct base_data result;
 575                         resolve_delta(child, base, &result);
 576                         if (i == ofs_last)
 577                                 free_base_data(base);
 578                         find_unresolved_deltas(&result, base);
 579                 }
 580         }
 581
 582         unlink_base_data(base);
 583 }
 584
 585 static int compare_delta_entry(const void *a, const void *b)
 586 {
 587         const struct delta_entry *delta_a = a;
 588         const struct delta_entry *delta_b = b;
 589         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 590 }
 591
 592 /* Parse all objects and return the pack content SHA1 hash */
 593 static void parse_pack_objects(unsigned char *sha1)
 594 {
 595         int i;
 596         struct delta_entry *delta = deltas;
 597         struct stat st;
 598
 599         /*
 600          * First pass:
 601          * - find locations of all objects;
 602          * - calculate SHA1 of all non-delta objects;
 603          * - remember base (SHA1 or offset) for all deltas.
 604          */
 605         if (verbose)
 606                 progress = start_progress(
 607                                 from_stdin ? "Receiving objects" : "Indexing objects",
 608                                 nr_objects);
 609         for (i = 0; i < nr_objects; i++) {
 610                 struct object_entry *obj = &objects[i];
 611                 void *data = unpack_raw_entry(obj, &delta->base);
 612                 obj->real_type = obj->type;
 613                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 614                         nr_deltas++;
 615                         delta->obj_no = i;
 616                         delta++;
 617                 } else
 618                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 619                 free(data);
 620                 display_progress(progress, i+1);
 621         }
 622         objects[i].idx.offset = consumed_bytes;
 623         stop_progress(&progress);
 624
 625         /* Check pack integrity */
 626         flush();
 627         git_SHA1_Final(sha1, &input_ctx);
 628         if (hashcmp(fill(20), sha1))
 629                 die("pack is corrupted (SHA1 mismatch)");
 630         use(20);
 631
 632         /* If input_fd is a file, we should have reached its end now. */
 633         if (fstat(input_fd, &st))
 634                 die_errno("cannot fstat packfile");
 635         if (S_ISREG(st.st_mode) &&
 636                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 637                 die("pack has junk at the end");
 638
 639         if (!nr_deltas)
 640                 return;
 641
 642         /* Sort deltas by base SHA1/offset for fast searching */
 643         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 644               compare_delta_entry);
 645
 646         /*
 647          * Second pass:
 648          * - for all non-delta objects, look if it is used as a base for
 649          *   deltas;
 650          * - if used as a base, uncompress the object and apply all deltas,
 651          *   recursively checking if the resulting object is used as a base
 652          *   for some more deltas.
 653          */
 654         if (verbose)
 655                 progress = start_progress("Resolving deltas", nr_deltas);
 656         for (i = 0; i < nr_objects; i++) {
 657                 struct object_entry *obj = &objects[i];
 658                 struct base_data base_obj;
 659
 660                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 661                         continue;
 662                 base_obj.obj = obj;
 663                 base_obj.data = NULL;
 664                 find_unresolved_deltas(&base_obj, NULL);
 665                 display_progress(progress, nr_resolved_deltas);
 666         }
 667 }
 668
 669 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 670 {
 671         z_stream stream;
 672         int status;
 673         unsigned char outbuf[4096];
 674
 675         memset(&stream, 0, sizeof(stream));
 676         deflateInit(&stream, zlib_compression_level);
 677         stream.next_in = in;
 678         stream.avail_in = size;
 679
 680         do {
 681                 stream.next_out = outbuf;
 682                 stream.avail_out = sizeof(outbuf);
 683                 status = deflate(&stream, Z_FINISH);
 684                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
 685         } while (status == Z_OK);
 686
 687         if (status != Z_STREAM_END)
 688                 die("unable to deflate appended object (%d)", status);
 689         size = stream.total_out;
 690         deflateEnd(&stream);
 691         return size;
 692 }
 693
 694 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 695                                const unsigned char *sha1, void *buf,
 696                                unsigned long size, enum object_type type)
 697 {
 698         struct object_entry *obj = &objects[nr_objects++];
 699         unsigned char header[10];
 700         unsigned long s = size;
 701         int n = 0;
 702         unsigned char c = (type << 4) | (s & 15);
 703         s >>= 4;
 704         while (s) {
 705                 header[n++] = c | 0x80;
 706                 c = s & 0x7f;
 707                 s >>= 7;
 708         }
 709         header[n++] = c;
 710         crc32_begin(f);
 711         sha1write(f, header, n);
 712         obj[0].size = size;
 713         obj[0].hdr_size = n;
 714         obj[0].type = type;
 715         obj[0].real_type = type;
 716         obj[1].idx.offset = obj[0].idx.offset + n;
 717         obj[1].idx.offset += write_compressed(f, buf, size);
 718         obj[0].idx.crc32 = crc32_end(f);
 719         sha1flush(f);
 720         hashcpy(obj->idx.sha1, sha1);
 721         return obj;
 722 }
 723
 724 static int delta_pos_compare(const void *_a, const void *_b)
 725 {
 726         struct delta_entry *a = *(struct delta_entry **)_a;
 727         struct delta_entry *b = *(struct delta_entry **)_b;
 728         return a->obj_no - b->obj_no;
 729 }
 730
 731 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 732 {
 733         struct delta_entry **sorted_by_pos;
 734         int i, n = 0;
 735
 736         /*
 737          * Since many unresolved deltas may well be themselves base objects
 738          * for more unresolved deltas, we really want to include the
 739          * smallest number of base objects that would cover as much delta
 740          * as possible by picking the
 741          * trunc deltas first, allowing for other deltas to resolve without
 742          * additional base objects.  Since most base objects are to be found
 743          * before deltas depending on them, a good heuristic is to start
 744          * resolving deltas in the same order as their position in the pack.
 745          */
 746         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 747         for (i = 0; i < nr_deltas; i++) {
 748                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 749                         continue;
 750                 sorted_by_pos[n++] = &deltas[i];
 751         }
 752         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 753
 754         for (i = 0; i < n; i++) {
 755                 struct delta_entry *d = sorted_by_pos[i];
 756                 enum object_type type;
 757                 struct base_data base_obj;
 758
 759                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 760                         continue;
 761                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 762                 if (!base_obj.data)
 763                         continue;
 764
 765                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 766                                 base_obj.size, typename(type)))
 767                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 768                 base_obj.obj = append_obj_to_pack(f, d->base.sha1,
 769                                         base_obj.data, base_obj.size, type);
 770                 find_unresolved_deltas(&base_obj, NULL);
 771                 display_progress(progress, nr_resolved_deltas);
 772         }
 773         free(sorted_by_pos);
 774 }
 775
 776 static void final(const char *final_pack_name, const char *curr_pack_name,
 777                   const char *final_index_name, const char *curr_index_name,
 778                   const char *keep_name, const char *keep_msg,
 779                   unsigned char *sha1)
 780 {
 781         const char *report = "pack";
 782         char name[PATH_MAX];
 783         int err;
 784
 785         if (!from_stdin) {
 786                 close(input_fd);
 787         } else {
 788                 fsync_or_die(output_fd, curr_pack_name);
 789                 err = close(output_fd);
 790                 if (err)
 791                         die_errno("error while closing pack file");
 792         }
 793
 794         if (keep_msg) {
 795                 int keep_fd, keep_msg_len = strlen(keep_msg);
 796
 797                 if (!keep_name)
 798                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
 799                 else
 800                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 801
 802                 if (keep_fd < 0) {
 803                         if (errno != EEXIST)
 804                                 die_errno("cannot write keep file '%s'",
 805                                           keep_name);
 806                 } else {
 807                         if (keep_msg_len > 0) {
 808                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 809                                 write_or_die(keep_fd, "\n", 1);
 810                         }
 811                         if (close(keep_fd) != 0)
 812                                 die_errno("cannot close written keep file '%s'",
 813                                     keep_name);
 814                         report = "keep";
 815                 }
 816         }
 817
 818         if (final_pack_name != curr_pack_name) {
 819                 if (!final_pack_name) {
 820                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 821                                  get_object_directory(), sha1_to_hex(sha1));
 822                         final_pack_name = name;
 823                 }
 824                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 825                         die("cannot store pack file");
 826         } else if (from_stdin)
 827                 chmod(final_pack_name, 0444);
 828
 829         if (final_index_name != curr_index_name) {
 830                 if (!final_index_name) {
 831                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 832                                  get_object_directory(), sha1_to_hex(sha1));
 833                         final_index_name = name;
 834                 }
 835                 if (move_temp_to_file(curr_index_name, final_index_name))
 836                         die("cannot store index file");
 837         } else
 838                 chmod(final_index_name, 0444);
 839
 840         if (!from_stdin) {
 841                 printf("%s\n", sha1_to_hex(sha1));
 842         } else {
 843                 char buf[48];
 844                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 845                                    report, sha1_to_hex(sha1));
 846                 write_or_die(1, buf, len);
 847
 848                 /*
 849                  * Let's just mimic git-unpack-objects here and write
 850                  * the last part of the input buffer to stdout.
 851                  */
 852                 while (input_len) {
 853                         err = xwrite(1, input_buffer + input_offset, input_len);
 854                         if (err <= 0)
 855                                 break;
 856                         input_len -= err;
 857                         input_offset += err;
 858                 }
 859         }
 860 }
 861
 862 static int git_index_pack_config(const char *k, const char *v, void *cb)
 863 {
 864         if (!strcmp(k, "pack.indexversion")) {
 865                 pack_idx_default_version = git_config_int(k, v);
 866                 if (pack_idx_default_version > 2)
 867                         die("bad pack.indexversion=%"PRIu32,
 868                                 pack_idx_default_version);
 869                 return 0;
 870         }
 871         return git_default_config(k, v, cb);
 872 }
 873
 874 int cmd_index_pack(int argc, const char **argv, const char *prefix)
 875 {
 876         int i, fix_thin_pack = 0;
 877         const char *curr_pack, *curr_index;
 878         const char *index_name = NULL, *pack_name = NULL;
 879         const char *keep_name = NULL, *keep_msg = NULL;
 880         char *index_name_buf = NULL, *keep_name_buf = NULL;
 881         struct pack_idx_entry **idx_objects;
 882         unsigned char pack_sha1[20];
 883         int nongit;
 884
 885         if (argc == 2 && !strcmp(argv[1], "-h"))
 886                 usage(index_pack_usage);
 887
 888         read_replace_refs = 0;
 889
 890         prefix = setup_git_directory_gently(&nongit);
 891         git_config(git_index_pack_config, NULL);
 892         if (prefix && chdir(prefix))
 893                 die("Cannot come back to cwd");
 894
 895         for (i = 1; i < argc; i++) {
 896                 const char *arg = argv[i];
 897
 898                 if (*arg == '-') {
 899                         if (!strcmp(arg, "--stdin")) {
 900                                 from_stdin = 1;
 901                         } else if (!strcmp(arg, "--fix-thin")) {
 902                                 fix_thin_pack = 1;
 903                         } else if (!strcmp(arg, "--strict")) {
 904                                 strict = 1;
 905                         } else if (!strcmp(arg, "--keep")) {
 906                                 keep_msg = "";
 907                         } else if (!prefixcmp(arg, "--keep=")) {
 908                                 keep_msg = arg + 7;
 909                         } else if (!prefixcmp(arg, "--pack_header=")) {
 910                                 struct pack_header *hdr;
 911                                 char *c;
 912
 913                                 hdr = (struct pack_header *)input_buffer;
 914                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 915                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 916                                 if (*c != ',')
 917                                         die("bad %s", arg);
 918                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 919                                 if (*c)
 920                                         die("bad %s", arg);
 921                                 input_len = sizeof(*hdr);
 922                         } else if (!strcmp(arg, "-v")) {
 923                                 verbose = 1;
 924                         } else if (!strcmp(arg, "-o")) {
 925                                 if (index_name || (i+1) >= argc)
 926                                         usage(index_pack_usage);
 927                                 index_name = argv[++i];
 928                         } else if (!prefixcmp(arg, "--index-version=")) {
 929                                 char *c;
 930                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 931                                 if (pack_idx_default_version > 2)
 932                                         die("bad %s", arg);
 933                                 if (*c == ',')
 934                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 935                                 if (*c || pack_idx_off32_limit & 0x80000000)
 936                                         die("bad %s", arg);
 937                         } else
 938                                 usage(index_pack_usage);
 939                         continue;
 940                 }
 941
 942                 if (pack_name)
 943                         usage(index_pack_usage);
 944                 pack_name = arg;
 945         }
 946
 947         if (!pack_name && !from_stdin)
 948                 usage(index_pack_usage);
 949         if (fix_thin_pack && !from_stdin)
 950                 die("--fix-thin cannot be used without --stdin");
 951         if (!index_name && pack_name) {
 952                 int len = strlen(pack_name);
 953                 if (!has_extension(pack_name, ".pack"))
 954                         die("packfile name '%s' does not end with '.pack'",
 955                             pack_name);
 956                 index_name_buf = xmalloc(len);
 957                 memcpy(index_name_buf, pack_name, len - 5);
 958                 strcpy(index_name_buf + len - 5, ".idx");
 959                 index_name = index_name_buf;
 960         }
 961         if (keep_msg && !keep_name && pack_name) {
 962                 int len = strlen(pack_name);
 963                 if (!has_extension(pack_name, ".pack"))
 964                         die("packfile name '%s' does not end with '.pack'",
 965                             pack_name);
 966                 keep_name_buf = xmalloc(len);
 967                 memcpy(keep_name_buf, pack_name, len - 5);
 968                 strcpy(keep_name_buf + len - 5, ".keep");
 969                 keep_name = keep_name_buf;
 970         }
 971
 972         curr_pack = open_pack_file(pack_name);
 973         parse_pack_header();
 974         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 975         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 976         parse_pack_objects(pack_sha1);
 977         if (nr_deltas == nr_resolved_deltas) {
 978                 stop_progress(&progress);
 979                 /* Flush remaining pack final 20-byte SHA1. */
 980                 flush();
 981         } else {
 982                 if (fix_thin_pack) {
 983                         struct sha1file *f;
 984                         unsigned char read_sha1[20], tail_sha1[20];
 985                         char msg[48];
 986                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 987                         int nr_objects_initial = nr_objects;
 988                         if (nr_unresolved <= 0)
 989                                 die("confusion beyond insanity");
 990                         objects = xrealloc(objects,
 991                                            (nr_objects + nr_unresolved + 1)
 992                                            * sizeof(*objects));
 993                         f = sha1fd(output_fd, curr_pack);
 994                         fix_unresolved_deltas(f, nr_unresolved);
 995                         sprintf(msg, "completed with %d local objects",
 996                                 nr_objects - nr_objects_initial);
 997                         stop_progress_msg(&progress, msg);
 998                         sha1close(f, tail_sha1, 0);
 999                         hashcpy(read_sha1, pack_sha1);
1000                         fixup_pack_header_footer(output_fd, pack_sha1,
1001                                                  curr_pack, nr_objects,
1002                                                  read_sha1, consumed_bytes-20);
1003                         if (hashcmp(read_sha1, tail_sha1) != 0)
1004                                 die("Unexpected tail checksum for %s "
1005                                     "(disk corruption?)", curr_pack);
1006                 }
1007                 if (nr_deltas != nr_resolved_deltas)
1008                         die("pack has %d unresolved deltas",
1009                             nr_deltas - nr_resolved_deltas);
1010         }
1011         free(deltas);
1012         if (strict)
1013                 check_objects();
1014
1015         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1016         for (i = 0; i < nr_objects; i++)
1017                 idx_objects[i] = &objects[i].idx;
1018         curr_index = write_idx_file(index_name, idx_objects, nr_objects, pack_sha1);
1019         free(idx_objects);
1020
1021         final(pack_name, curr_pack,
1022                 index_name, curr_index,
1023                 keep_name, keep_msg,
1024                 pack_sha1);
1025         free(objects);
1026         free(index_name_buf);
1027         free(keep_name_buf);
1028         if (pack_name == NULL)
1029                 free((void *) curr_pack);
1030         if (index_name == NULL)
1031                 free((void *) curr_index);
1032
1033         return 0;
1034 }