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                         output_fd = odb_mkstemp(tmpfile, sizeof(tmpfile),
 175                                                 "pack/tmp_pack_XXXXXX");
 176                         pack_name = xstrdup(tmpfile);
 177                 } else
 178                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 179                 if (output_fd < 0)
 180                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 181                 pack_fd = output_fd;
 182         } else {
 183                 input_fd = open(pack_name, O_RDONLY);
 184                 if (input_fd < 0)
 185                         die("cannot open packfile '%s': %s",
 186                             pack_name, strerror(errno));
 187                 output_fd = -1;
 188                 pack_fd = input_fd;
 189         }
 190         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 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                 if (!n)
 370                         die("premature end of pack file, %lu bytes missing",
 371                             len - rdy);
 372                 rdy += n;
 373         } while (rdy < len);
 374         data = xmalloc(obj->size);
 375         memset(&stream, 0, sizeof(stream));
 376         stream.next_out = data;
 377         stream.avail_out = obj->size;
 378         stream.next_in = src;
 379         stream.avail_in = len;
 380         inflateInit(&stream);
 381         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 382         inflateEnd(&stream);
 383         if (st != Z_STREAM_END || stream.total_out != obj->size)
 384                 die("serious inflate inconsistency");
 385         free(src);
 386         return data;
 387 }
 388
 389 static int find_delta(const union delta_base *base)
 390 {
 391         int first = 0, last = nr_deltas;
 392
 393         while (first < last) {
 394                 int next = (first + last) / 2;
 395                 struct delta_entry *delta = &deltas[next];
 396                 int cmp;
 397
 398                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 399                 if (!cmp)
 400                         return next;
 401                 if (cmp < 0) {
 402                         last = next;
 403                         continue;
 404                 }
 405                 first = next+1;
 406         }
 407         return -first-1;
 408 }
 409
 410 static int find_delta_children(const union delta_base *base,
 411                                int *first_index, int *last_index)
 412 {
 413         int first = find_delta(base);
 414         int last = first;
 415         int end = nr_deltas - 1;
 416
 417         if (first < 0)
 418                 return -1;
 419         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 420                 --first;
 421         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 422                 ++last;
 423         *first_index = first;
 424         *last_index = last;
 425         return 0;
 426 }
 427
 428 static void sha1_object(const void *data, unsigned long size,
 429                         enum object_type type, unsigned char *sha1)
 430 {
 431         hash_sha1_file(data, size, typename(type), sha1);
 432         if (has_sha1_file(sha1)) {
 433                 void *has_data;
 434                 enum object_type has_type;
 435                 unsigned long has_size;
 436                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 437                 if (!has_data)
 438                         die("cannot read existing object %s", sha1_to_hex(sha1));
 439                 if (size != has_size || type != has_type ||
 440                     memcmp(data, has_data, size) != 0)
 441                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 442                 free(has_data);
 443         }
 444         if (strict) {
 445                 if (type == OBJ_BLOB) {
 446                         struct blob *blob = lookup_blob(sha1);
 447                         if (blob)
 448                                 blob->object.flags |= FLAG_CHECKED;
 449                         else
 450                                 die("invalid blob object %s", sha1_to_hex(sha1));
 451                 } else {
 452                         struct object *obj;
 453                         int eaten;
 454                         void *buf = (void *) data;
 455
 456                         /*
 457                          * we do not need to free the memory here, as the
 458                          * buf is deleted by the caller.
 459                          */
 460                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 461                         if (!obj)
 462                                 die("invalid %s", typename(type));
 463                         if (fsck_object(obj, 1, fsck_error_function))
 464                                 die("Error in object");
 465                         if (fsck_walk(obj, mark_link, 0))
 466                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 467
 468                         if (obj->type == OBJ_TREE) {
 469                                 struct tree *item = (struct tree *) obj;
 470                                 item->buffer = NULL;
 471                         }
 472                         if (obj->type == OBJ_COMMIT) {
 473                                 struct commit *commit = (struct commit *) obj;
 474                                 commit->buffer = NULL;
 475                         }
 476                         obj->flags |= FLAG_CHECKED;
 477                 }
 478         }
 479 }
 480
 481 static void *get_base_data(struct base_data *c)
 482 {
 483         if (!c->data) {
 484                 struct object_entry *obj = c->obj;
 485
 486                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 487                         void *base = get_base_data(c->base);
 488                         void *raw = get_data_from_pack(obj);
 489                         c->data = patch_delta(
 490                                 base, c->base->size,
 491                                 raw, obj->size,
 492                                 &c->size);
 493                         free(raw);
 494                         if (!c->data)
 495                                 bad_object(obj->idx.offset, "failed to apply delta");
 496                 } else
 497                         c->data = get_data_from_pack(obj);
 498
 499                 base_cache_used += c->size;
 500                 prune_base_data(c);
 501         }
 502         return c->data;
 503 }
 504
 505 static void resolve_delta(struct object_entry *delta_obj,
 506                           struct base_data *base_obj, enum object_type type)
 507 {
 508         void *delta_data;
 509         unsigned long delta_size;
 510         union delta_base delta_base;
 511         int j, first, last;
 512         struct base_data result;
 513
 514         delta_obj->real_type = type;
 515         delta_data = get_data_from_pack(delta_obj);
 516         delta_size = delta_obj->size;
 517         result.data = patch_delta(get_base_data(base_obj), base_obj->size,
 518                              delta_data, delta_size,
 519                              &result.size);
 520         free(delta_data);
 521         if (!result.data)
 522                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 523         sha1_object(result.data, result.size, type, delta_obj->idx.sha1);
 524         nr_resolved_deltas++;
 525
 526         result.obj = delta_obj;
 527         link_base_data(base_obj, &result);
 528
 529         hashcpy(delta_base.sha1, delta_obj->idx.sha1);
 530         if (!find_delta_children(&delta_base, &first, &last)) {
 531                 for (j = first; j <= last; j++) {
 532                         struct object_entry *child = objects + deltas[j].obj_no;
 533                         if (child->real_type == OBJ_REF_DELTA)
 534                                 resolve_delta(child, &result, type);
 535                 }
 536         }
 537
 538         memset(&delta_base, 0, sizeof(delta_base));
 539         delta_base.offset = delta_obj->idx.offset;
 540         if (!find_delta_children(&delta_base, &first, &last)) {
 541                 for (j = first; j <= last; j++) {
 542                         struct object_entry *child = objects + deltas[j].obj_no;
 543                         if (child->real_type == OBJ_OFS_DELTA)
 544                                 resolve_delta(child, &result, type);
 545                 }
 546         }
 547
 548         unlink_base_data(&result);
 549 }
 550
 551 static int compare_delta_entry(const void *a, const void *b)
 552 {
 553         const struct delta_entry *delta_a = a;
 554         const struct delta_entry *delta_b = b;
 555         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 556 }
 557
 558 /* Parse all objects and return the pack content SHA1 hash */
 559 static void parse_pack_objects(unsigned char *sha1)
 560 {
 561         int i;
 562         struct delta_entry *delta = deltas;
 563         struct stat st;
 564
 565         /*
 566          * First pass:
 567          * - find locations of all objects;
 568          * - calculate SHA1 of all non-delta objects;
 569          * - remember base (SHA1 or offset) for all deltas.
 570          */
 571         if (verbose)
 572                 progress = start_progress(
 573                                 from_stdin ? "Receiving objects" : "Indexing objects",
 574                                 nr_objects);
 575         for (i = 0; i < nr_objects; i++) {
 576                 struct object_entry *obj = &objects[i];
 577                 void *data = unpack_raw_entry(obj, &delta->base);
 578                 obj->real_type = obj->type;
 579                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 580                         nr_deltas++;
 581                         delta->obj_no = i;
 582                         delta++;
 583                 } else
 584                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 585                 free(data);
 586                 display_progress(progress, i+1);
 587         }
 588         objects[i].idx.offset = consumed_bytes;
 589         stop_progress(&progress);
 590
 591         /* Check pack integrity */
 592         flush();
 593         SHA1_Final(sha1, &input_ctx);
 594         if (hashcmp(fill(20), sha1))
 595                 die("pack is corrupted (SHA1 mismatch)");
 596         use(20);
 597
 598         /* If input_fd is a file, we should have reached its end now. */
 599         if (fstat(input_fd, &st))
 600                 die("cannot fstat packfile: %s", strerror(errno));
 601         if (S_ISREG(st.st_mode) &&
 602                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 603                 die("pack has junk at the end");
 604
 605         if (!nr_deltas)
 606                 return;
 607
 608         /* Sort deltas by base SHA1/offset for fast searching */
 609         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 610               compare_delta_entry);
 611
 612         /*
 613          * Second pass:
 614          * - for all non-delta objects, look if it is used as a base for
 615          *   deltas;
 616          * - if used as a base, uncompress the object and apply all deltas,
 617          *   recursively checking if the resulting object is used as a base
 618          *   for some more deltas.
 619          */
 620         if (verbose)
 621                 progress = start_progress("Resolving deltas", nr_deltas);
 622         for (i = 0; i < nr_objects; i++) {
 623                 struct object_entry *obj = &objects[i];
 624                 union delta_base base;
 625                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 626                 struct base_data base_obj;
 627
 628                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 629                         continue;
 630                 hashcpy(base.sha1, obj->idx.sha1);
 631                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 632                 memset(&base, 0, sizeof(base));
 633                 base.offset = obj->idx.offset;
 634                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 635                 if (!ref && !ofs)
 636                         continue;
 637                 base_obj.data = get_data_from_pack(obj);
 638                 base_obj.size = obj->size;
 639                 base_obj.obj = obj;
 640                 link_base_data(NULL, &base_obj);
 641
 642                 if (ref)
 643                         for (j = ref_first; j <= ref_last; j++) {
 644                                 struct object_entry *child = objects + deltas[j].obj_no;
 645                                 if (child->real_type == OBJ_REF_DELTA)
 646                                         resolve_delta(child, &base_obj, obj->type);
 647                         }
 648                 if (ofs)
 649                         for (j = ofs_first; j <= ofs_last; j++) {
 650                                 struct object_entry *child = objects + deltas[j].obj_no;
 651                                 if (child->real_type == OBJ_OFS_DELTA)
 652                                         resolve_delta(child, &base_obj, obj->type);
 653                         }
 654                 unlink_base_data(&base_obj);
 655                 display_progress(progress, nr_resolved_deltas);
 656         }
 657 }
 658
 659 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 660 {
 661         z_stream stream;
 662         unsigned long maxsize;
 663         void *out;
 664
 665         memset(&stream, 0, sizeof(stream));
 666         deflateInit(&stream, zlib_compression_level);
 667         maxsize = deflateBound(&stream, size);
 668         out = xmalloc(maxsize);
 669
 670         /* Compress it */
 671         stream.next_in = in;
 672         stream.avail_in = size;
 673         stream.next_out = out;
 674         stream.avail_out = maxsize;
 675         while (deflate(&stream, Z_FINISH) == Z_OK);
 676         deflateEnd(&stream);
 677
 678         size = stream.total_out;
 679         sha1write(f, out, size);
 680         free(out);
 681         return size;
 682 }
 683
 684 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 685                                const unsigned char *sha1, void *buf,
 686                                unsigned long size, enum object_type type)
 687 {
 688         struct object_entry *obj = &objects[nr_objects++];
 689         unsigned char header[10];
 690         unsigned long s = size;
 691         int n = 0;
 692         unsigned char c = (type << 4) | (s & 15);
 693         s >>= 4;
 694         while (s) {
 695                 header[n++] = c | 0x80;
 696                 c = s & 0x7f;
 697                 s >>= 7;
 698         }
 699         header[n++] = c;
 700         crc32_begin(f);
 701         sha1write(f, header, n);
 702         obj[0].size = size;
 703         obj[0].hdr_size = n;
 704         obj[0].type = type;
 705         obj[0].real_type = type;
 706         obj[1].idx.offset = obj[0].idx.offset + n;
 707         obj[1].idx.offset += write_compressed(f, buf, size);
 708         obj[0].idx.crc32 = crc32_end(f);
 709         sha1flush(f);
 710         hashcpy(obj->idx.sha1, sha1);
 711         return obj;
 712 }
 713
 714 static int delta_pos_compare(const void *_a, const void *_b)
 715 {
 716         struct delta_entry *a = *(struct delta_entry **)_a;
 717         struct delta_entry *b = *(struct delta_entry **)_b;
 718         return a->obj_no - b->obj_no;
 719 }
 720
 721 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 722 {
 723         struct delta_entry **sorted_by_pos;
 724         int i, n = 0;
 725
 726         /*
 727          * Since many unresolved deltas may well be themselves base objects
 728          * for more unresolved deltas, we really want to include the
 729          * smallest number of base objects that would cover as much delta
 730          * as possible by picking the
 731          * trunc deltas first, allowing for other deltas to resolve without
 732          * additional base objects.  Since most base objects are to be found
 733          * before deltas depending on them, a good heuristic is to start
 734          * resolving deltas in the same order as their position in the pack.
 735          */
 736         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 737         for (i = 0; i < nr_deltas; i++) {
 738                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 739                         continue;
 740                 sorted_by_pos[n++] = &deltas[i];
 741         }
 742         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 743
 744         for (i = 0; i < n; i++) {
 745                 struct delta_entry *d = sorted_by_pos[i];
 746                 enum object_type type;
 747                 int j, first, last;
 748                 struct base_data base_obj;
 749
 750                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 751                         continue;
 752                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 753                 if (!base_obj.data)
 754                         continue;
 755
 756                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 757                                 base_obj.size, typename(type)))
 758                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 759                 base_obj.obj = append_obj_to_pack(f, d->base.sha1,
 760                                         base_obj.data, base_obj.size, type);
 761                 link_base_data(NULL, &base_obj);
 762
 763                 find_delta_children(&d->base, &first, &last);
 764                 for (j = first; j <= last; j++) {
 765                         struct object_entry *child = objects + deltas[j].obj_no;
 766                         if (child->real_type == OBJ_REF_DELTA)
 767                                 resolve_delta(child, &base_obj, type);
 768                 }
 769
 770                 unlink_base_data(&base_obj);
 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("error while closing pack file: %s", strerror(errno));
 792                 chmod(curr_pack_name, 0444);
 793         }
 794
 795         if (keep_msg) {
 796                 int keep_fd, keep_msg_len = strlen(keep_msg);
 797
 798                 if (!keep_name)
 799                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
 800                 else
 801                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 802
 803                 if (keep_fd < 0) {
 804                         if (errno != EEXIST)
 805                                 die("cannot write keep file '%s' (%s)",
 806                                     keep_name, strerror(errno));
 807                 } else {
 808                         if (keep_msg_len > 0) {
 809                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 810                                 write_or_die(keep_fd, "\n", 1);
 811                         }
 812                         if (close(keep_fd) != 0)
 813                                 die("cannot close written keep file '%s' (%s)",
 814                                     keep_name, strerror(errno));
 815                         report = "keep";
 816                 }
 817         }
 818
 819         if (final_pack_name != curr_pack_name) {
 820                 if (!final_pack_name) {
 821                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 822                                  get_object_directory(), sha1_to_hex(sha1));
 823                         final_pack_name = name;
 824                 }
 825                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 826                         die("cannot store pack file");
 827         }
 828
 829         chmod(curr_index_name, 0444);
 830         if (final_index_name != curr_index_name) {
 831                 if (!final_index_name) {
 832                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 833                                  get_object_directory(), sha1_to_hex(sha1));
 834                         final_index_name = name;
 835                 }
 836                 if (move_temp_to_file(curr_index_name, final_index_name))
 837                         die("cannot store index file");
 838         }
 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 main(int argc, char **argv)
 875 {
 876         int i, fix_thin_pack = 0;
 877         char *curr_pack, *pack_name = NULL;
 878         char *curr_index, *index_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
 884         /*
 885          * We wish to read the repository's config file if any, and
 886          * for that it is necessary to call setup_git_directory_gently().
 887          * However if the cwd was inside .git/objects/pack/ then we need
 888          * to go back there or all the pack name arguments will be wrong.
 889          * And in that case we cannot rely on any prefix returned by
 890          * setup_git_directory_gently() either.
 891          */
 892         {
 893                 char cwd[PATH_MAX+1];
 894                 int nongit;
 895
 896                 if (!getcwd(cwd, sizeof(cwd)-1))
 897                         die("Unable to get current working directory");
 898                 setup_git_directory_gently(&nongit);
 899                 git_config(git_index_pack_config, NULL);
 900                 if (chdir(cwd))
 901                         die("Cannot come back to cwd");
 902         }
 903
 904         for (i = 1; i < argc; i++) {
 905                 char *arg = argv[i];
 906
 907                 if (*arg == '-') {
 908                         if (!strcmp(arg, "--stdin")) {
 909                                 from_stdin = 1;
 910                         } else if (!strcmp(arg, "--fix-thin")) {
 911                                 fix_thin_pack = 1;
 912                         } else if (!strcmp(arg, "--strict")) {
 913                                 strict = 1;
 914                         } else if (!strcmp(arg, "--keep")) {
 915                                 keep_msg = "";
 916                         } else if (!prefixcmp(arg, "--keep=")) {
 917                                 keep_msg = arg + 7;
 918                         } else if (!prefixcmp(arg, "--pack_header=")) {
 919                                 struct pack_header *hdr;
 920                                 char *c;
 921
 922                                 hdr = (struct pack_header *)input_buffer;
 923                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 924                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 925                                 if (*c != ',')
 926                                         die("bad %s", arg);
 927                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 928                                 if (*c)
 929                                         die("bad %s", arg);
 930                                 input_len = sizeof(*hdr);
 931                         } else if (!strcmp(arg, "-v")) {
 932                                 verbose = 1;
 933                         } else if (!strcmp(arg, "-o")) {
 934                                 if (index_name || (i+1) >= argc)
 935                                         usage(index_pack_usage);
 936                                 index_name = argv[++i];
 937                         } else if (!prefixcmp(arg, "--index-version=")) {
 938                                 char *c;
 939                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 940                                 if (pack_idx_default_version > 2)
 941                                         die("bad %s", arg);
 942                                 if (*c == ',')
 943                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 944                                 if (*c || pack_idx_off32_limit & 0x80000000)
 945                                         die("bad %s", arg);
 946                         } else
 947                                 usage(index_pack_usage);
 948                         continue;
 949                 }
 950
 951                 if (pack_name)
 952                         usage(index_pack_usage);
 953                 pack_name = arg;
 954         }
 955
 956         if (!pack_name && !from_stdin)
 957                 usage(index_pack_usage);
 958         if (fix_thin_pack && !from_stdin)
 959                 die("--fix-thin cannot be used without --stdin");
 960         if (!index_name && pack_name) {
 961                 int len = strlen(pack_name);
 962                 if (!has_extension(pack_name, ".pack"))
 963                         die("packfile name '%s' does not end with '.pack'",
 964                             pack_name);
 965                 index_name_buf = xmalloc(len);
 966                 memcpy(index_name_buf, pack_name, len - 5);
 967                 strcpy(index_name_buf + len - 5, ".idx");
 968                 index_name = index_name_buf;
 969         }
 970         if (keep_msg && !keep_name && pack_name) {
 971                 int len = strlen(pack_name);
 972                 if (!has_extension(pack_name, ".pack"))
 973                         die("packfile name '%s' does not end with '.pack'",
 974                             pack_name);
 975                 keep_name_buf = xmalloc(len);
 976                 memcpy(keep_name_buf, pack_name, len - 5);
 977                 strcpy(keep_name_buf + len - 5, ".keep");
 978                 keep_name = keep_name_buf;
 979         }
 980
 981         curr_pack = open_pack_file(pack_name);
 982         parse_pack_header();
 983         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 984         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 985         parse_pack_objects(pack_sha1);
 986         if (nr_deltas == nr_resolved_deltas) {
 987                 stop_progress(&progress);
 988                 /* Flush remaining pack final 20-byte SHA1. */
 989                 flush();
 990         } else {
 991                 if (fix_thin_pack) {
 992                         struct sha1file *f;
 993                         unsigned char read_sha1[20], tail_sha1[20];
 994                         char msg[48];
 995                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 996                         int nr_objects_initial = nr_objects;
 997                         if (nr_unresolved <= 0)
 998                                 die("confusion beyond insanity");
 999                         objects = xrealloc(objects,
1000                                            (nr_objects + nr_unresolved + 1)
1001                                            * sizeof(*objects));
1002                         f = sha1fd(output_fd, curr_pack);
1003                         fix_unresolved_deltas(f, nr_unresolved);
1004                         sprintf(msg, "completed with %d local objects",
1005                                 nr_objects - nr_objects_initial);
1006                         stop_progress_msg(&progress, msg);
1007                         sha1close(f, tail_sha1, 0);
1008                         hashcpy(read_sha1, pack_sha1);
1009                         fixup_pack_header_footer(output_fd, pack_sha1,
1010                                                  curr_pack, nr_objects,
1011                                                  read_sha1, consumed_bytes-20);
1012                         if (hashcmp(read_sha1, tail_sha1) != 0)
1013                                 die("Unexpected tail checksum for %s "
1014                                     "(disk corruption?)", curr_pack);
1015                 }
1016                 if (nr_deltas != nr_resolved_deltas)
1017                         die("pack has %d unresolved deltas",
1018                             nr_deltas - nr_resolved_deltas);
1019         }
1020         free(deltas);
1021         if (strict)
1022                 check_objects();
1023
1024         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1025         for (i = 0; i < nr_objects; i++)
1026                 idx_objects[i] = &objects[i].idx;
1027         curr_index = write_idx_file(index_name, idx_objects, nr_objects, pack_sha1);
1028         free(idx_objects);
1029
1030         final(pack_name, curr_pack,
1031                 index_name, curr_index,
1032                 keep_name, keep_msg,
1033                 pack_sha1);
1034         free(objects);
1035         free(index_name_buf);
1036         free(keep_name_buf);
1037         if (pack_name == NULL)
1038                 free(curr_pack);
1039         if (index_name == NULL)
1040                 free(curr_index);
1041
1042         return 0;
1043 }