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
  12 static const char index_pack_usage[] =
  13 "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] [--strict] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  14
  15 struct object_entry
  16 {
  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 {
  48         union delta_base base;
  49         int obj_no;
  50 };
  51
  52 static struct object_entry *objects;
  53 static struct delta_entry *deltas;
  54 static struct base_data *base_cache;
  55 static size_t base_cache_used;
  56 static int nr_objects;
  57 static int nr_deltas;
  58 static int nr_resolved_deltas;
  59
  60 static int from_stdin;
  61 static int strict;
  62 static int verbose;
  63
  64 static struct progress *progress;
  65
  66 /* We always read in 4kB chunks. */
  67 static unsigned char input_buffer[4096];
  68 static unsigned int input_offset, input_len;
  69 static off_t consumed_bytes;
  70 static SHA_CTX input_ctx;
  71 static uint32_t input_crc32;
  72 static int input_fd, output_fd, pack_fd;
  73
  74 static int mark_link(struct object *obj, int type, void *data)
  75 {
  76         if (!obj)
  77                 return -1;
  78
  79         if (type != OBJ_ANY && obj->type != type)
  80                 die("object type mismatch at %s", sha1_to_hex(obj->sha1));
  81
  82         obj->flags |= FLAG_LINK;
  83         return 0;
  84 }
  85
  86 /* The content of each linked object must have been checked
  87    or it must be already present in the object database */
  88 static void check_object(struct object *obj)
  89 {
  90         if (!obj)
  91                 return;
  92
  93         if (!(obj->flags & FLAG_LINK))
  94                 return;
  95
  96         if (!(obj->flags & FLAG_CHECKED)) {
  97                 unsigned long size;
  98                 int type = sha1_object_info(obj->sha1, &size);
  99                 if (type != obj->type || type <= 0)
 100                         die("object of unexpected type");
 101                 obj->flags |= FLAG_CHECKED;
 102                 return;
 103         }
 104 }
 105
 106 static void check_objects(void)
 107 {
 108         unsigned i, max;
 109
 110         max = get_max_object_index();
 111         for (i = 0; i < max; i++)
 112                 check_object(get_indexed_object(i));
 113 }
 114
 115
 116 /* Discard current buffer used content. */
 117 static void flush(void)
 118 {
 119         if (input_offset) {
 120                 if (output_fd >= 0)
 121                         write_or_die(output_fd, input_buffer, input_offset);
 122                 SHA1_Update(&input_ctx, input_buffer, input_offset);
 123                 memmove(input_buffer, input_buffer + input_offset, input_len);
 124                 input_offset = 0;
 125         }
 126 }
 127
 128 /*
 129  * Make sure at least "min" bytes are available in the buffer, and
 130  * return the pointer to the buffer.
 131  */
 132 static void *fill(int min)
 133 {
 134         if (min <= input_len)
 135                 return input_buffer + input_offset;
 136         if (min > sizeof(input_buffer))
 137                 die("cannot fill %d bytes", min);
 138         flush();
 139         do {
 140                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 141                                 sizeof(input_buffer) - input_len);
 142                 if (ret <= 0) {
 143                         if (!ret)
 144                                 die("early EOF");
 145                         die("read error on input: %s", strerror(errno));
 146                 }
 147                 input_len += ret;
 148                 if (from_stdin)
 149                         display_throughput(progress, consumed_bytes + input_len);
 150         } while (input_len < min);
 151         return input_buffer;
 152 }
 153
 154 static void use(int bytes)
 155 {
 156         if (bytes > input_len)
 157                 die("used more bytes than were available");
 158         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 159         input_len -= bytes;
 160         input_offset += bytes;
 161
 162         /* make sure off_t is sufficiently large not to wrap */
 163         if (consumed_bytes > consumed_bytes + bytes)
 164                 die("pack too large for current definition of off_t");
 165         consumed_bytes += bytes;
 166 }
 167
 168 static char *open_pack_file(char *pack_name)
 169 {
 170         if (from_stdin) {
 171                 input_fd = 0;
 172                 if (!pack_name) {
 173                         static char tmpfile[PATH_MAX];
 174                         snprintf(tmpfile, sizeof(tmpfile),
 175                                  "%s/tmp_pack_XXXXXX", get_object_directory());
 176                         output_fd = xmkstemp(tmpfile);
 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("unable to create %s: %s\n", pack_name, strerror(errno));
 182                 pack_fd = output_fd;
 183         } else {
 184                 input_fd = open(pack_name, O_RDONLY);
 185                 if (input_fd < 0)
 186                         die("cannot open packfile '%s': %s",
 187                             pack_name, strerror(errno));
 188                 output_fd = -1;
 189                 pack_fd = input_fd;
 190         }
 191         SHA1_Init(&input_ctx);
 192         return pack_name;
 193 }
 194
 195 static void parse_pack_header(void)
 196 {
 197         struct pack_header *hdr = fill(sizeof(struct pack_header));
 198
 199         /* Header consistency check */
 200         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 201                 die("pack signature mismatch");
 202         if (!pack_version_ok(hdr->hdr_version))
 203                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 204
 205         nr_objects = ntohl(hdr->hdr_entries);
 206         use(sizeof(struct pack_header));
 207 }
 208
 209 static void bad_object(unsigned long offset, const char *format,
 210                        ...) NORETURN __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 prune_base_data(struct base_data *retain)
 224 {
 225         struct base_data *b = base_cache;
 226         for (b = base_cache;
 227              base_cache_used > delta_base_cache_limit && b;
 228              b = b->child) {
 229                 if (b->data && b != retain) {
 230                         free(b->data);
 231                         b->data = NULL;
 232                         base_cache_used -= b->size;
 233                 }
 234         }
 235 }
 236
 237 static void link_base_data(struct base_data *base, struct base_data *c)
 238 {
 239         if (base)
 240                 base->child = c;
 241         else
 242                 base_cache = c;
 243
 244         c->base = base;
 245         c->child = NULL;
 246         base_cache_used += c->size;
 247         prune_base_data(c);
 248 }
 249
 250 static void unlink_base_data(struct base_data *c)
 251 {
 252         struct base_data *base = c->base;
 253         if (base)
 254                 base->child = NULL;
 255         else
 256                 base_cache = NULL;
 257         if (c->data) {
 258                 free(c->data);
 259                 base_cache_used -= c->size;
 260         }
 261 }
 262
 263 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 264 {
 265         z_stream stream;
 266         void *buf = xmalloc(size);
 267
 268         memset(&stream, 0, sizeof(stream));
 269         stream.next_out = buf;
 270         stream.avail_out = size;
 271         stream.next_in = fill(1);
 272         stream.avail_in = input_len;
 273         inflateInit(&stream);
 274
 275         for (;;) {
 276                 int ret = inflate(&stream, 0);
 277                 use(input_len - stream.avail_in);
 278                 if (stream.total_out == size && ret == Z_STREAM_END)
 279                         break;
 280                 if (ret != Z_OK)
 281                         bad_object(offset, "inflate returned %d", ret);
 282                 stream.next_in = fill(1);
 283                 stream.avail_in = input_len;
 284         }
 285         inflateEnd(&stream);
 286         return buf;
 287 }
 288
 289 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 290 {
 291         unsigned char *p, c;
 292         unsigned long size;
 293         off_t base_offset;
 294         unsigned shift;
 295         void *data;
 296
 297         obj->idx.offset = consumed_bytes;
 298         input_crc32 = crc32(0, Z_NULL, 0);
 299
 300         p = fill(1);
 301         c = *p;
 302         use(1);
 303         obj->type = (c >> 4) & 7;
 304         size = (c & 15);
 305         shift = 4;
 306         while (c & 0x80) {
 307                 p = fill(1);
 308                 c = *p;
 309                 use(1);
 310                 size += (c & 0x7fUL) << shift;
 311                 shift += 7;
 312         }
 313         obj->size = size;
 314
 315         switch (obj->type) {
 316         case OBJ_REF_DELTA:
 317                 hashcpy(delta_base->sha1, fill(20));
 318                 use(20);
 319                 break;
 320         case OBJ_OFS_DELTA:
 321                 memset(delta_base, 0, sizeof(*delta_base));
 322                 p = fill(1);
 323                 c = *p;
 324                 use(1);
 325                 base_offset = c & 127;
 326                 while (c & 128) {
 327                         base_offset += 1;
 328                         if (!base_offset || MSB(base_offset, 7))
 329                                 bad_object(obj->idx.offset, "offset value overflow for delta base object");
 330                         p = fill(1);
 331                         c = *p;
 332                         use(1);
 333                         base_offset = (base_offset << 7) + (c & 127);
 334                 }
 335                 delta_base->offset = obj->idx.offset - base_offset;
 336                 if (delta_base->offset >= obj->idx.offset)
 337                         bad_object(obj->idx.offset, "delta base offset is out of bound");
 338                 break;
 339         case OBJ_COMMIT:
 340         case OBJ_TREE:
 341         case OBJ_BLOB:
 342         case OBJ_TAG:
 343                 break;
 344         default:
 345                 bad_object(obj->idx.offset, "unknown object type %d", obj->type);
 346         }
 347         obj->hdr_size = consumed_bytes - obj->idx.offset;
 348
 349         data = unpack_entry_data(obj->idx.offset, obj->size);
 350         obj->idx.crc32 = input_crc32;
 351         return data;
 352 }
 353
 354 static void *get_data_from_pack(struct object_entry *obj)
 355 {
 356         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 357         unsigned long len = obj[1].idx.offset - from;
 358         unsigned long rdy = 0;
 359         unsigned char *src, *data;
 360         z_stream stream;
 361         int st;
 362
 363         src = xmalloc(len);
 364         data = src;
 365         do {
 366                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 367                 if (n <= 0)
 368                         die("cannot pread pack file: %s", strerror(errno));
 369                 rdy += n;
 370         } while (rdy < len);
 371         data = xmalloc(obj->size);
 372         memset(&stream, 0, sizeof(stream));
 373         stream.next_out = data;
 374         stream.avail_out = obj->size;
 375         stream.next_in = src;
 376         stream.avail_in = len;
 377         inflateInit(&stream);
 378         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 379         inflateEnd(&stream);
 380         if (st != Z_STREAM_END || stream.total_out != obj->size)
 381                 die("serious inflate inconsistency");
 382         free(src);
 383         return data;
 384 }
 385
 386 static int find_delta(const union delta_base *base)
 387 {
 388         int first = 0, last = nr_deltas;
 389
 390         while (first < last) {
 391                 int next = (first + last) / 2;
 392                 struct delta_entry *delta = &deltas[next];
 393                 int cmp;
 394
 395                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 396                 if (!cmp)
 397                         return next;
 398                 if (cmp < 0) {
 399                         last = next;
 400                         continue;
 401                 }
 402                 first = next+1;
 403         }
 404         return -first-1;
 405 }
 406
 407 static int find_delta_children(const union delta_base *base,
 408                                int *first_index, int *last_index)
 409 {
 410         int first = find_delta(base);
 411         int last = first;
 412         int end = nr_deltas - 1;
 413
 414         if (first < 0)
 415                 return -1;
 416         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 417                 --first;
 418         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 419                 ++last;
 420         *first_index = first;
 421         *last_index = last;
 422         return 0;
 423 }
 424
 425 static void sha1_object(const void *data, unsigned long size,
 426                         enum object_type type, unsigned char *sha1)
 427 {
 428         hash_sha1_file(data, size, typename(type), sha1);
 429         if (has_sha1_file(sha1)) {
 430                 void *has_data;
 431                 enum object_type has_type;
 432                 unsigned long has_size;
 433                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 434                 if (!has_data)
 435                         die("cannot read existing object %s", sha1_to_hex(sha1));
 436                 if (size != has_size || type != has_type ||
 437                     memcmp(data, has_data, size) != 0)
 438                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 439                 free(has_data);
 440         }
 441         if (strict) {
 442                 if (type == OBJ_BLOB) {
 443                         struct blob *blob = lookup_blob(sha1);
 444                         if (blob)
 445                                 blob->object.flags |= FLAG_CHECKED;
 446                         else
 447                                 die("invalid blob object %s", sha1_to_hex(sha1));
 448                 } else {
 449                         struct object *obj;
 450                         int eaten;
 451                         void *buf = (void *) data;
 452
 453                         /*
 454                          * we do not need to free the memory here, as the
 455                          * buf is deleted by the caller.
 456                          */
 457                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 458                         if (!obj)
 459                                 die("invalid %s", typename(type));
 460                         if (fsck_object(obj, 1, fsck_error_function))
 461                                 die("Error in object");
 462                         if (fsck_walk(obj, mark_link, 0))
 463                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 464
 465                         if (obj->type == OBJ_TREE) {
 466                                 struct tree *item = (struct tree *) obj;
 467                                 item->buffer = NULL;
 468                         }
 469                         if (obj->type == OBJ_COMMIT) {
 470                                 struct commit *commit = (struct commit *) obj;
 471                                 commit->buffer = NULL;
 472                         }
 473                         obj->flags |= FLAG_CHECKED;
 474                 }
 475         }
 476 }
 477
 478 static void *get_base_data(struct base_data *c)
 479 {
 480         if (!c->data) {
 481                 struct object_entry *obj = c->obj;
 482
 483                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 484                         void *base = get_base_data(c->base);
 485                         void *raw = get_data_from_pack(obj);
 486                         c->data = patch_delta(
 487                                 base, c->base->size,
 488                                 raw, obj->size,
 489                                 &c->size);
 490                         free(raw);
 491                         if (!c->data)
 492                                 bad_object(obj->idx.offset, "failed to apply delta");
 493                 } else
 494                         c->data = get_data_from_pack(obj);
 495
 496                 base_cache_used += c->size;
 497                 prune_base_data(c);
 498         }
 499         return c->data;
 500 }
 501
 502 static void resolve_delta(struct object_entry *delta_obj,
 503                           struct base_data *base_obj, enum object_type type)
 504 {
 505         void *delta_data;
 506         unsigned long delta_size;
 507         union delta_base delta_base;
 508         int j, first, last;
 509         struct base_data result;
 510
 511         delta_obj->real_type = type;
 512         delta_data = get_data_from_pack(delta_obj);
 513         delta_size = delta_obj->size;
 514         result.data = patch_delta(get_base_data(base_obj), base_obj->size,
 515                              delta_data, delta_size,
 516                              &result.size);
 517         free(delta_data);
 518         if (!result.data)
 519                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 520         sha1_object(result.data, result.size, type, delta_obj->idx.sha1);
 521         nr_resolved_deltas++;
 522
 523         result.obj = delta_obj;
 524         link_base_data(base_obj, &result);
 525
 526         hashcpy(delta_base.sha1, delta_obj->idx.sha1);
 527         if (!find_delta_children(&delta_base, &first, &last)) {
 528                 for (j = first; j <= last; j++) {
 529                         struct object_entry *child = objects + deltas[j].obj_no;
 530                         if (child->real_type == OBJ_REF_DELTA)
 531                                 resolve_delta(child, &result, type);
 532                 }
 533         }
 534
 535         memset(&delta_base, 0, sizeof(delta_base));
 536         delta_base.offset = delta_obj->idx.offset;
 537         if (!find_delta_children(&delta_base, &first, &last)) {
 538                 for (j = first; j <= last; j++) {
 539                         struct object_entry *child = objects + deltas[j].obj_no;
 540                         if (child->real_type == OBJ_OFS_DELTA)
 541                                 resolve_delta(child, &result, type);
 542                 }
 543         }
 544
 545         unlink_base_data(&result);
 546 }
 547
 548 static int compare_delta_entry(const void *a, const void *b)
 549 {
 550         const struct delta_entry *delta_a = a;
 551         const struct delta_entry *delta_b = b;
 552         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 553 }
 554
 555 /* Parse all objects and return the pack content SHA1 hash */
 556 static void parse_pack_objects(unsigned char *sha1)
 557 {
 558         int i;
 559         struct delta_entry *delta = deltas;
 560         struct stat st;
 561
 562         /*
 563          * First pass:
 564          * - find locations of all objects;
 565          * - calculate SHA1 of all non-delta objects;
 566          * - remember base (SHA1 or offset) for all deltas.
 567          */
 568         if (verbose)
 569                 progress = start_progress(
 570                                 from_stdin ? "Receiving objects" : "Indexing objects",
 571                                 nr_objects);
 572         for (i = 0; i < nr_objects; i++) {
 573                 struct object_entry *obj = &objects[i];
 574                 void *data = unpack_raw_entry(obj, &delta->base);
 575                 obj->real_type = obj->type;
 576                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 577                         nr_deltas++;
 578                         delta->obj_no = i;
 579                         delta++;
 580                 } else
 581                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 582                 free(data);
 583                 display_progress(progress, i+1);
 584         }
 585         objects[i].idx.offset = consumed_bytes;
 586         stop_progress(&progress);
 587
 588         /* Check pack integrity */
 589         flush();
 590         SHA1_Final(sha1, &input_ctx);
 591         if (hashcmp(fill(20), sha1))
 592                 die("pack is corrupted (SHA1 mismatch)");
 593         use(20);
 594
 595         /* If input_fd is a file, we should have reached its end now. */
 596         if (fstat(input_fd, &st))
 597                 die("cannot fstat packfile: %s", strerror(errno));
 598         if (S_ISREG(st.st_mode) &&
 599                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 600                 die("pack has junk at the end");
 601
 602         if (!nr_deltas)
 603                 return;
 604
 605         /* Sort deltas by base SHA1/offset for fast searching */
 606         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 607               compare_delta_entry);
 608
 609         /*
 610          * Second pass:
 611          * - for all non-delta objects, look if it is used as a base for
 612          *   deltas;
 613          * - if used as a base, uncompress the object and apply all deltas,
 614          *   recursively checking if the resulting object is used as a base
 615          *   for some more deltas.
 616          */
 617         if (verbose)
 618                 progress = start_progress("Resolving deltas", nr_deltas);
 619         for (i = 0; i < nr_objects; i++) {
 620                 struct object_entry *obj = &objects[i];
 621                 union delta_base base;
 622                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 623                 struct base_data base_obj;
 624
 625                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 626                         continue;
 627                 hashcpy(base.sha1, obj->idx.sha1);
 628                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 629                 memset(&base, 0, sizeof(base));
 630                 base.offset = obj->idx.offset;
 631                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 632                 if (!ref && !ofs)
 633                         continue;
 634                 base_obj.data = get_data_from_pack(obj);
 635                 base_obj.size = obj->size;
 636                 base_obj.obj = obj;
 637                 link_base_data(NULL, &base_obj);
 638
 639                 if (ref)
 640                         for (j = ref_first; j <= ref_last; j++) {
 641                                 struct object_entry *child = objects + deltas[j].obj_no;
 642                                 if (child->real_type == OBJ_REF_DELTA)
 643                                         resolve_delta(child, &base_obj, obj->type);
 644                         }
 645                 if (ofs)
 646                         for (j = ofs_first; j <= ofs_last; j++) {
 647                                 struct object_entry *child = objects + deltas[j].obj_no;
 648                                 if (child->real_type == OBJ_OFS_DELTA)
 649                                         resolve_delta(child, &base_obj, obj->type);
 650                         }
 651                 unlink_base_data(&base_obj);
 652                 display_progress(progress, nr_resolved_deltas);
 653         }
 654 }
 655
 656 static int write_compressed(int fd, void *in, unsigned int size, uint32_t *obj_crc)
 657 {
 658         z_stream stream;
 659         unsigned long maxsize;
 660         void *out;
 661
 662         memset(&stream, 0, sizeof(stream));
 663         deflateInit(&stream, zlib_compression_level);
 664         maxsize = deflateBound(&stream, size);
 665         out = xmalloc(maxsize);
 666
 667         /* Compress it */
 668         stream.next_in = in;
 669         stream.avail_in = size;
 670         stream.next_out = out;
 671         stream.avail_out = maxsize;
 672         while (deflate(&stream, Z_FINISH) == Z_OK);
 673         deflateEnd(&stream);
 674
 675         size = stream.total_out;
 676         write_or_die(fd, out, size);
 677         *obj_crc = crc32(*obj_crc, out, size);
 678         free(out);
 679         return size;
 680 }
 681
 682 static struct object_entry *append_obj_to_pack(
 683                                const unsigned char *sha1, void *buf,
 684                                unsigned long size, enum object_type type)
 685 {
 686         struct object_entry *obj = &objects[nr_objects++];
 687         unsigned char header[10];
 688         unsigned long s = size;
 689         int n = 0;
 690         unsigned char c = (type << 4) | (s & 15);
 691         s >>= 4;
 692         while (s) {
 693                 header[n++] = c | 0x80;
 694                 c = s & 0x7f;
 695                 s >>= 7;
 696         }
 697         header[n++] = c;
 698         write_or_die(output_fd, header, n);
 699         obj[0].idx.crc32 = crc32(0, Z_NULL, 0);
 700         obj[0].idx.crc32 = crc32(obj[0].idx.crc32, header, n);
 701         obj[1].idx.offset = obj[0].idx.offset + n;
 702         obj[1].idx.offset += write_compressed(output_fd, buf, size, &obj[0].idx.crc32);
 703         hashcpy(obj->idx.sha1, sha1);
 704         return obj;
 705 }
 706
 707 static int delta_pos_compare(const void *_a, const void *_b)
 708 {
 709         struct delta_entry *a = *(struct delta_entry **)_a;
 710         struct delta_entry *b = *(struct delta_entry **)_b;
 711         return a->obj_no - b->obj_no;
 712 }
 713
 714 static void fix_unresolved_deltas(int nr_unresolved)
 715 {
 716         struct delta_entry **sorted_by_pos;
 717         int i, n = 0;
 718
 719         /*
 720          * Since many unresolved deltas may well be themselves base objects
 721          * for more unresolved deltas, we really want to include the
 722          * smallest number of base objects that would cover as much delta
 723          * as possible by picking the
 724          * trunc deltas first, allowing for other deltas to resolve without
 725          * additional base objects.  Since most base objects are to be found
 726          * before deltas depending on them, a good heuristic is to start
 727          * resolving deltas in the same order as their position in the pack.
 728          */
 729         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 730         for (i = 0; i < nr_deltas; i++) {
 731                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 732                         continue;
 733                 sorted_by_pos[n++] = &deltas[i];
 734         }
 735         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 736
 737         for (i = 0; i < n; i++) {
 738                 struct delta_entry *d = sorted_by_pos[i];
 739                 enum object_type type;
 740                 int j, first, last;
 741                 struct base_data base_obj;
 742
 743                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 744                         continue;
 745                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 746                 if (!base_obj.data)
 747                         continue;
 748
 749                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 750                                 base_obj.size, typename(type)))
 751                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 752                 base_obj.obj = append_obj_to_pack(d->base.sha1, base_obj.data,
 753                         base_obj.size, type);
 754                 link_base_data(NULL, &base_obj);
 755
 756                 find_delta_children(&d->base, &first, &last);
 757                 for (j = first; j <= last; j++) {
 758                         struct object_entry *child = objects + deltas[j].obj_no;
 759                         if (child->real_type == OBJ_REF_DELTA)
 760                                 resolve_delta(child, &base_obj, type);
 761                 }
 762
 763                 unlink_base_data(&base_obj);
 764                 display_progress(progress, nr_resolved_deltas);
 765         }
 766         free(sorted_by_pos);
 767 }
 768
 769 static void final(const char *final_pack_name, const char *curr_pack_name,
 770                   const char *final_index_name, const char *curr_index_name,
 771                   const char *keep_name, const char *keep_msg,
 772                   unsigned char *sha1)
 773 {
 774         const char *report = "pack";
 775         char name[PATH_MAX];
 776         int err;
 777
 778         if (!from_stdin) {
 779                 close(input_fd);
 780         } else {
 781                 fsync_or_die(output_fd, curr_pack_name);
 782                 err = close(output_fd);
 783                 if (err)
 784                         die("error while closing pack file: %s", strerror(errno));
 785                 chmod(curr_pack_name, 0444);
 786         }
 787
 788         if (keep_msg) {
 789                 int keep_fd, keep_msg_len = strlen(keep_msg);
 790                 if (!keep_name) {
 791                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 792                                  get_object_directory(), sha1_to_hex(sha1));
 793                         keep_name = name;
 794                 }
 795                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 796                 if (keep_fd < 0) {
 797                         if (errno != EEXIST)
 798                                 die("cannot write keep file");
 799                 } else {
 800                         if (keep_msg_len > 0) {
 801                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 802                                 write_or_die(keep_fd, "\n", 1);
 803                         }
 804                         if (close(keep_fd) != 0)
 805                                 die("cannot write keep file");
 806                         report = "keep";
 807                 }
 808         }
 809
 810         if (final_pack_name != curr_pack_name) {
 811                 if (!final_pack_name) {
 812                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 813                                  get_object_directory(), sha1_to_hex(sha1));
 814                         final_pack_name = name;
 815                 }
 816                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 817                         die("cannot store pack file");
 818         }
 819
 820         chmod(curr_index_name, 0444);
 821         if (final_index_name != curr_index_name) {
 822                 if (!final_index_name) {
 823                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 824                                  get_object_directory(), sha1_to_hex(sha1));
 825                         final_index_name = name;
 826                 }
 827                 if (move_temp_to_file(curr_index_name, final_index_name))
 828                         die("cannot store index file");
 829         }
 830
 831         if (!from_stdin) {
 832                 printf("%s\n", sha1_to_hex(sha1));
 833         } else {
 834                 char buf[48];
 835                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 836                                    report, sha1_to_hex(sha1));
 837                 write_or_die(1, buf, len);
 838
 839                 /*
 840                  * Let's just mimic git-unpack-objects here and write
 841                  * the last part of the input buffer to stdout.
 842                  */
 843                 while (input_len) {
 844                         err = xwrite(1, input_buffer + input_offset, input_len);
 845                         if (err <= 0)
 846                                 break;
 847                         input_len -= err;
 848                         input_offset += err;
 849                 }
 850         }
 851 }
 852
 853 static int git_index_pack_config(const char *k, const char *v, void *cb)
 854 {
 855         if (!strcmp(k, "pack.indexversion")) {
 856                 pack_idx_default_version = git_config_int(k, v);
 857                 if (pack_idx_default_version > 2)
 858                         die("bad pack.indexversion=%d", pack_idx_default_version);
 859                 return 0;
 860         }
 861         return git_default_config(k, v, cb);
 862 }
 863
 864 int main(int argc, char **argv)
 865 {
 866         int i, fix_thin_pack = 0;
 867         char *curr_pack, *pack_name = NULL;
 868         char *curr_index, *index_name = NULL;
 869         const char *keep_name = NULL, *keep_msg = NULL;
 870         char *index_name_buf = NULL, *keep_name_buf = NULL;
 871         struct pack_idx_entry **idx_objects;
 872         unsigned char sha1[20];
 873
 874         git_config(git_index_pack_config, NULL);
 875
 876         for (i = 1; i < argc; i++) {
 877                 char *arg = argv[i];
 878
 879                 if (*arg == '-') {
 880                         if (!strcmp(arg, "--stdin")) {
 881                                 from_stdin = 1;
 882                         } else if (!strcmp(arg, "--fix-thin")) {
 883                                 fix_thin_pack = 1;
 884                         } else if (!strcmp(arg, "--strict")) {
 885                                 strict = 1;
 886                         } else if (!strcmp(arg, "--keep")) {
 887                                 keep_msg = "";
 888                         } else if (!prefixcmp(arg, "--keep=")) {
 889                                 keep_msg = arg + 7;
 890                         } else if (!prefixcmp(arg, "--pack_header=")) {
 891                                 struct pack_header *hdr;
 892                                 char *c;
 893
 894                                 hdr = (struct pack_header *)input_buffer;
 895                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 896                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 897                                 if (*c != ',')
 898                                         die("bad %s", arg);
 899                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 900                                 if (*c)
 901                                         die("bad %s", arg);
 902                                 input_len = sizeof(*hdr);
 903                         } else if (!strcmp(arg, "-v")) {
 904                                 verbose = 1;
 905                         } else if (!strcmp(arg, "-o")) {
 906                                 if (index_name || (i+1) >= argc)
 907                                         usage(index_pack_usage);
 908                                 index_name = argv[++i];
 909                         } else if (!prefixcmp(arg, "--index-version=")) {
 910                                 char *c;
 911                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 912                                 if (pack_idx_default_version > 2)
 913                                         die("bad %s", arg);
 914                                 if (*c == ',')
 915                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 916                                 if (*c || pack_idx_off32_limit & 0x80000000)
 917                                         die("bad %s", arg);
 918                         } else
 919                                 usage(index_pack_usage);
 920                         continue;
 921                 }
 922
 923                 if (pack_name)
 924                         usage(index_pack_usage);
 925                 pack_name = arg;
 926         }
 927
 928         if (!pack_name && !from_stdin)
 929                 usage(index_pack_usage);
 930         if (fix_thin_pack && !from_stdin)
 931                 die("--fix-thin cannot be used without --stdin");
 932         if (!index_name && pack_name) {
 933                 int len = strlen(pack_name);
 934                 if (!has_extension(pack_name, ".pack"))
 935                         die("packfile name '%s' does not end with '.pack'",
 936                             pack_name);
 937                 index_name_buf = xmalloc(len);
 938                 memcpy(index_name_buf, pack_name, len - 5);
 939                 strcpy(index_name_buf + len - 5, ".idx");
 940                 index_name = index_name_buf;
 941         }
 942         if (keep_msg && !keep_name && pack_name) {
 943                 int len = strlen(pack_name);
 944                 if (!has_extension(pack_name, ".pack"))
 945                         die("packfile name '%s' does not end with '.pack'",
 946                             pack_name);
 947                 keep_name_buf = xmalloc(len);
 948                 memcpy(keep_name_buf, pack_name, len - 5);
 949                 strcpy(keep_name_buf + len - 5, ".keep");
 950                 keep_name = keep_name_buf;
 951         }
 952
 953         curr_pack = open_pack_file(pack_name);
 954         parse_pack_header();
 955         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 956         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 957         parse_pack_objects(sha1);
 958         if (nr_deltas == nr_resolved_deltas) {
 959                 stop_progress(&progress);
 960                 /* Flush remaining pack final 20-byte SHA1. */
 961                 flush();
 962         } else {
 963                 if (fix_thin_pack) {
 964                         char msg[48];
 965                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 966                         int nr_objects_initial = nr_objects;
 967                         if (nr_unresolved <= 0)
 968                                 die("confusion beyond insanity");
 969                         objects = xrealloc(objects,
 970                                            (nr_objects + nr_unresolved + 1)
 971                                            * sizeof(*objects));
 972                         fix_unresolved_deltas(nr_unresolved);
 973                         sprintf(msg, "completed with %d local objects",
 974                                 nr_objects - nr_objects_initial);
 975                         stop_progress_msg(&progress, msg);
 976                         fixup_pack_header_footer(output_fd, sha1,
 977                                                  curr_pack, nr_objects);
 978                 }
 979                 if (nr_deltas != nr_resolved_deltas)
 980                         die("pack has %d unresolved deltas",
 981                             nr_deltas - nr_resolved_deltas);
 982         }
 983         free(deltas);
 984         if (strict)
 985                 check_objects();
 986
 987         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
 988         for (i = 0; i < nr_objects; i++)
 989                 idx_objects[i] = &objects[i].idx;
 990         curr_index = write_idx_file(index_name, idx_objects, nr_objects, sha1);
 991         free(idx_objects);
 992
 993         final(pack_name, curr_pack,
 994                 index_name, curr_index,
 995                 keep_name, keep_msg,
 996                 sha1);
 997         free(objects);
 998         free(index_name_buf);
 999         free(keep_name_buf);
1000         if (pack_name == NULL)
1001                 free(curr_pack);
1002         if (index_name == NULL)
1003                 free(curr_index);
1004
1005         return 0;
1006 }