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/pack/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 %"PRIu32" unsupported",
 204                         ntohl(hdr->hdr_version));
 205
 206         nr_objects = ntohl(hdr->hdr_entries);
 207         use(sizeof(struct pack_header));
 208 }
 209
 210 static void bad_object(unsigned long offset, const char *format,
 211                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 212
 213 static void bad_object(unsigned long offset, const char *format, ...)
 214 {
 215         va_list params;
 216         char buf[1024];
 217
 218         va_start(params, format);
 219         vsnprintf(buf, sizeof(buf), format, params);
 220         va_end(params);
 221         die("pack has bad object at offset %lu: %s", offset, buf);
 222 }
 223
 224 static void prune_base_data(struct base_data *retain)
 225 {
 226         struct base_data *b = base_cache;
 227         for (b = base_cache;
 228              base_cache_used > delta_base_cache_limit && b;
 229              b = b->child) {
 230                 if (b->data && b != retain) {
 231                         free(b->data);
 232                         b->data = NULL;
 233                         base_cache_used -= b->size;
 234                 }
 235         }
 236 }
 237
 238 static void link_base_data(struct base_data *base, struct base_data *c)
 239 {
 240         if (base)
 241                 base->child = c;
 242         else
 243                 base_cache = c;
 244
 245         c->base = base;
 246         c->child = NULL;
 247         base_cache_used += c->size;
 248         prune_base_data(c);
 249 }
 250
 251 static void unlink_base_data(struct base_data *c)
 252 {
 253         struct base_data *base = c->base;
 254         if (base)
 255                 base->child = NULL;
 256         else
 257                 base_cache = NULL;
 258         if (c->data) {
 259                 free(c->data);
 260                 base_cache_used -= c->size;
 261         }
 262 }
 263
 264 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 265 {
 266         z_stream stream;
 267         void *buf = xmalloc(size);
 268
 269         memset(&stream, 0, sizeof(stream));
 270         stream.next_out = buf;
 271         stream.avail_out = size;
 272         stream.next_in = fill(1);
 273         stream.avail_in = input_len;
 274         inflateInit(&stream);
 275
 276         for (;;) {
 277                 int ret = inflate(&stream, 0);
 278                 use(input_len - stream.avail_in);
 279                 if (stream.total_out == size && ret == Z_STREAM_END)
 280                         break;
 281                 if (ret != Z_OK)
 282                         bad_object(offset, "inflate returned %d", ret);
 283                 stream.next_in = fill(1);
 284                 stream.avail_in = input_len;
 285         }
 286         inflateEnd(&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, c;
 293         unsigned long size;
 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 & 0x7fUL) << 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 >= 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 long rdy = 0;
 360         unsigned char *src, *data;
 361         z_stream stream;
 362         int st;
 363
 364         src = xmalloc(len);
 365         data = src;
 366         do {
 367                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 368                 if (n < 0)
 369                         die("cannot pread pack file: %s", strerror(errno));
 370                 if (!n)
 371                         die("premature end of pack file, %lu bytes missing",
 372                             len - rdy);
 373                 rdy += n;
 374         } while (rdy < len);
 375         data = xmalloc(obj->size);
 376         memset(&stream, 0, sizeof(stream));
 377         stream.next_out = data;
 378         stream.avail_out = obj->size;
 379         stream.next_in = src;
 380         stream.avail_in = len;
 381         inflateInit(&stream);
 382         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 383         inflateEnd(&stream);
 384         if (st != Z_STREAM_END || stream.total_out != obj->size)
 385                 die("serious inflate inconsistency");
 386         free(src);
 387         return data;
 388 }
 389
 390 static int find_delta(const union delta_base *base)
 391 {
 392         int first = 0, last = nr_deltas;
 393
 394         while (first < last) {
 395                 int next = (first + last) / 2;
 396                 struct delta_entry *delta = &deltas[next];
 397                 int cmp;
 398
 399                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 400                 if (!cmp)
 401                         return next;
 402                 if (cmp < 0) {
 403                         last = next;
 404                         continue;
 405                 }
 406                 first = next+1;
 407         }
 408         return -first-1;
 409 }
 410
 411 static int find_delta_children(const union delta_base *base,
 412                                int *first_index, int *last_index)
 413 {
 414         int first = find_delta(base);
 415         int last = first;
 416         int end = nr_deltas - 1;
 417
 418         if (first < 0)
 419                 return -1;
 420         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 421                 --first;
 422         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 423                 ++last;
 424         *first_index = first;
 425         *last_index = last;
 426         return 0;
 427 }
 428
 429 static void sha1_object(const void *data, unsigned long size,
 430                         enum object_type type, unsigned char *sha1)
 431 {
 432         hash_sha1_file(data, size, typename(type), sha1);
 433         if (has_sha1_file(sha1)) {
 434                 void *has_data;
 435                 enum object_type has_type;
 436                 unsigned long has_size;
 437                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 438                 if (!has_data)
 439                         die("cannot read existing object %s", sha1_to_hex(sha1));
 440                 if (size != has_size || type != has_type ||
 441                     memcmp(data, has_data, size) != 0)
 442                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 443                 free(has_data);
 444         }
 445         if (strict) {
 446                 if (type == OBJ_BLOB) {
 447                         struct blob *blob = lookup_blob(sha1);
 448                         if (blob)
 449                                 blob->object.flags |= FLAG_CHECKED;
 450                         else
 451                                 die("invalid blob object %s", sha1_to_hex(sha1));
 452                 } else {
 453                         struct object *obj;
 454                         int eaten;
 455                         void *buf = (void *) data;
 456
 457                         /*
 458                          * we do not need to free the memory here, as the
 459                          * buf is deleted by the caller.
 460                          */
 461                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 462                         if (!obj)
 463                                 die("invalid %s", typename(type));
 464                         if (fsck_object(obj, 1, fsck_error_function))
 465                                 die("Error in object");
 466                         if (fsck_walk(obj, mark_link, 0))
 467                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 468
 469                         if (obj->type == OBJ_TREE) {
 470                                 struct tree *item = (struct tree *) obj;
 471                                 item->buffer = NULL;
 472                         }
 473                         if (obj->type == OBJ_COMMIT) {
 474                                 struct commit *commit = (struct commit *) obj;
 475                                 commit->buffer = NULL;
 476                         }
 477                         obj->flags |= FLAG_CHECKED;
 478                 }
 479         }
 480 }
 481
 482 static void *get_base_data(struct base_data *c)
 483 {
 484         if (!c->data) {
 485                 struct object_entry *obj = c->obj;
 486
 487                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 488                         void *base = get_base_data(c->base);
 489                         void *raw = get_data_from_pack(obj);
 490                         c->data = patch_delta(
 491                                 base, c->base->size,
 492                                 raw, obj->size,
 493                                 &c->size);
 494                         free(raw);
 495                         if (!c->data)
 496                                 bad_object(obj->idx.offset, "failed to apply delta");
 497                 } else
 498                         c->data = get_data_from_pack(obj);
 499
 500                 base_cache_used += c->size;
 501                 prune_base_data(c);
 502         }
 503         return c->data;
 504 }
 505
 506 static void resolve_delta(struct object_entry *delta_obj,
 507                           struct base_data *base_obj, enum object_type type)
 508 {
 509         void *delta_data;
 510         unsigned long delta_size;
 511         union delta_base delta_base;
 512         int j, first, last;
 513         struct base_data result;
 514
 515         delta_obj->real_type = type;
 516         delta_data = get_data_from_pack(delta_obj);
 517         delta_size = delta_obj->size;
 518         result.data = patch_delta(get_base_data(base_obj), base_obj->size,
 519                              delta_data, delta_size,
 520                              &result.size);
 521         free(delta_data);
 522         if (!result.data)
 523                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 524         sha1_object(result.data, result.size, type, delta_obj->idx.sha1);
 525         nr_resolved_deltas++;
 526
 527         result.obj = delta_obj;
 528         link_base_data(base_obj, &result);
 529
 530         hashcpy(delta_base.sha1, delta_obj->idx.sha1);
 531         if (!find_delta_children(&delta_base, &first, &last)) {
 532                 for (j = first; j <= last; j++) {
 533                         struct object_entry *child = objects + deltas[j].obj_no;
 534                         if (child->real_type == OBJ_REF_DELTA)
 535                                 resolve_delta(child, &result, type);
 536                 }
 537         }
 538
 539         memset(&delta_base, 0, sizeof(delta_base));
 540         delta_base.offset = delta_obj->idx.offset;
 541         if (!find_delta_children(&delta_base, &first, &last)) {
 542                 for (j = first; j <= last; j++) {
 543                         struct object_entry *child = objects + deltas[j].obj_no;
 544                         if (child->real_type == OBJ_OFS_DELTA)
 545                                 resolve_delta(child, &result, type);
 546                 }
 547         }
 548
 549         unlink_base_data(&result);
 550 }
 551
 552 static int compare_delta_entry(const void *a, const void *b)
 553 {
 554         const struct delta_entry *delta_a = a;
 555         const struct delta_entry *delta_b = b;
 556         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 557 }
 558
 559 /* Parse all objects and return the pack content SHA1 hash */
 560 static void parse_pack_objects(unsigned char *sha1)
 561 {
 562         int i;
 563         struct delta_entry *delta = deltas;
 564         struct stat st;
 565
 566         /*
 567          * First pass:
 568          * - find locations of all objects;
 569          * - calculate SHA1 of all non-delta objects;
 570          * - remember base (SHA1 or offset) for all deltas.
 571          */
 572         if (verbose)
 573                 progress = start_progress(
 574                                 from_stdin ? "Receiving objects" : "Indexing objects",
 575                                 nr_objects);
 576         for (i = 0; i < nr_objects; i++) {
 577                 struct object_entry *obj = &objects[i];
 578                 void *data = unpack_raw_entry(obj, &delta->base);
 579                 obj->real_type = obj->type;
 580                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 581                         nr_deltas++;
 582                         delta->obj_no = i;
 583                         delta++;
 584                 } else
 585                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 586                 free(data);
 587                 display_progress(progress, i+1);
 588         }
 589         objects[i].idx.offset = consumed_bytes;
 590         stop_progress(&progress);
 591
 592         /* Check pack integrity */
 593         flush();
 594         SHA1_Final(sha1, &input_ctx);
 595         if (hashcmp(fill(20), sha1))
 596                 die("pack is corrupted (SHA1 mismatch)");
 597         use(20);
 598
 599         /* If input_fd is a file, we should have reached its end now. */
 600         if (fstat(input_fd, &st))
 601                 die("cannot fstat packfile: %s", strerror(errno));
 602         if (S_ISREG(st.st_mode) &&
 603                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 604                 die("pack has junk at the end");
 605
 606         if (!nr_deltas)
 607                 return;
 608
 609         /* Sort deltas by base SHA1/offset for fast searching */
 610         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 611               compare_delta_entry);
 612
 613         /*
 614          * Second pass:
 615          * - for all non-delta objects, look if it is used as a base for
 616          *   deltas;
 617          * - if used as a base, uncompress the object and apply all deltas,
 618          *   recursively checking if the resulting object is used as a base
 619          *   for some more deltas.
 620          */
 621         if (verbose)
 622                 progress = start_progress("Resolving deltas", nr_deltas);
 623         for (i = 0; i < nr_objects; i++) {
 624                 struct object_entry *obj = &objects[i];
 625                 union delta_base base;
 626                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 627                 struct base_data base_obj;
 628
 629                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 630                         continue;
 631                 hashcpy(base.sha1, obj->idx.sha1);
 632                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 633                 memset(&base, 0, sizeof(base));
 634                 base.offset = obj->idx.offset;
 635                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 636                 if (!ref && !ofs)
 637                         continue;
 638                 base_obj.data = get_data_from_pack(obj);
 639                 base_obj.size = obj->size;
 640                 base_obj.obj = obj;
 641                 link_base_data(NULL, &base_obj);
 642
 643                 if (ref)
 644                         for (j = ref_first; j <= ref_last; j++) {
 645                                 struct object_entry *child = objects + deltas[j].obj_no;
 646                                 if (child->real_type == OBJ_REF_DELTA)
 647                                         resolve_delta(child, &base_obj, obj->type);
 648                         }
 649                 if (ofs)
 650                         for (j = ofs_first; j <= ofs_last; j++) {
 651                                 struct object_entry *child = objects + deltas[j].obj_no;
 652                                 if (child->real_type == OBJ_OFS_DELTA)
 653                                         resolve_delta(child, &base_obj, obj->type);
 654                         }
 655                 unlink_base_data(&base_obj);
 656                 display_progress(progress, nr_resolved_deltas);
 657         }
 658 }
 659
 660 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 661 {
 662         z_stream stream;
 663         unsigned long maxsize;
 664         void *out;
 665
 666         memset(&stream, 0, sizeof(stream));
 667         deflateInit(&stream, zlib_compression_level);
 668         maxsize = deflateBound(&stream, size);
 669         out = xmalloc(maxsize);
 670
 671         /* Compress it */
 672         stream.next_in = in;
 673         stream.avail_in = size;
 674         stream.next_out = out;
 675         stream.avail_out = maxsize;
 676         while (deflate(&stream, Z_FINISH) == Z_OK);
 677         deflateEnd(&stream);
 678
 679         size = stream.total_out;
 680         sha1write(f, out, size);
 681         free(out);
 682         return size;
 683 }
 684
 685 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 686                                const unsigned char *sha1, void *buf,
 687                                unsigned long size, enum object_type type)
 688 {
 689         struct object_entry *obj = &objects[nr_objects++];
 690         unsigned char header[10];
 691         unsigned long s = size;
 692         int n = 0;
 693         unsigned char c = (type << 4) | (s & 15);
 694         s >>= 4;
 695         while (s) {
 696                 header[n++] = c | 0x80;
 697                 c = s & 0x7f;
 698                 s >>= 7;
 699         }
 700         header[n++] = c;
 701         crc32_begin(f);
 702         sha1write(f, header, n);
 703         obj[0].size = size;
 704         obj[0].hdr_size = n;
 705         obj[0].type = type;
 706         obj[0].real_type = type;
 707         obj[1].idx.offset = obj[0].idx.offset + n;
 708         obj[1].idx.offset += write_compressed(f, buf, size);
 709         obj[0].idx.crc32 = crc32_end(f);
 710         sha1flush(f);
 711         hashcpy(obj->idx.sha1, sha1);
 712         return obj;
 713 }
 714
 715 static int delta_pos_compare(const void *_a, const void *_b)
 716 {
 717         struct delta_entry *a = *(struct delta_entry **)_a;
 718         struct delta_entry *b = *(struct delta_entry **)_b;
 719         return a->obj_no - b->obj_no;
 720 }
 721
 722 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 723 {
 724         struct delta_entry **sorted_by_pos;
 725         int i, n = 0;
 726
 727         /*
 728          * Since many unresolved deltas may well be themselves base objects
 729          * for more unresolved deltas, we really want to include the
 730          * smallest number of base objects that would cover as much delta
 731          * as possible by picking the
 732          * trunc deltas first, allowing for other deltas to resolve without
 733          * additional base objects.  Since most base objects are to be found
 734          * before deltas depending on them, a good heuristic is to start
 735          * resolving deltas in the same order as their position in the pack.
 736          */
 737         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 738         for (i = 0; i < nr_deltas; i++) {
 739                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 740                         continue;
 741                 sorted_by_pos[n++] = &deltas[i];
 742         }
 743         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 744
 745         for (i = 0; i < n; i++) {
 746                 struct delta_entry *d = sorted_by_pos[i];
 747                 enum object_type type;
 748                 int j, first, last;
 749                 struct base_data base_obj;
 750
 751                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 752                         continue;
 753                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 754                 if (!base_obj.data)
 755                         continue;
 756
 757                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 758                                 base_obj.size, typename(type)))
 759                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 760                 base_obj.obj = append_obj_to_pack(f, d->base.sha1,
 761                                         base_obj.data, base_obj.size, type);
 762                 link_base_data(NULL, &base_obj);
 763
 764                 find_delta_children(&d->base, &first, &last);
 765                 for (j = first; j <= last; j++) {
 766                         struct object_entry *child = objects + deltas[j].obj_no;
 767                         if (child->real_type == OBJ_REF_DELTA)
 768                                 resolve_delta(child, &base_obj, type);
 769                 }
 770
 771                 unlink_base_data(&base_obj);
 772                 display_progress(progress, nr_resolved_deltas);
 773         }
 774         free(sorted_by_pos);
 775 }
 776
 777 static void final(const char *final_pack_name, const char *curr_pack_name,
 778                   const char *final_index_name, const char *curr_index_name,
 779                   const char *keep_name, const char *keep_msg,
 780                   unsigned char *sha1)
 781 {
 782         const char *report = "pack";
 783         char name[PATH_MAX];
 784         int err;
 785
 786         if (!from_stdin) {
 787                 close(input_fd);
 788         } else {
 789                 fsync_or_die(output_fd, curr_pack_name);
 790                 err = close(output_fd);
 791                 if (err)
 792                         die("error while closing pack file: %s", strerror(errno));
 793                 chmod(curr_pack_name, 0444);
 794         }
 795
 796         if (keep_msg) {
 797                 int keep_fd, keep_msg_len = strlen(keep_msg);
 798                 if (!keep_name) {
 799                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 800                                  get_object_directory(), sha1_to_hex(sha1));
 801                         keep_name = name;
 802                 }
 803                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 804                 if (keep_fd < 0) {
 805                         if (errno != EEXIST)
 806                                 die("cannot write keep file");
 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 write keep file");
 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         }
 827
 828         chmod(curr_index_name, 0444);
 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         }
 838
 839         if (!from_stdin) {
 840                 printf("%s\n", sha1_to_hex(sha1));
 841         } else {
 842                 char buf[48];
 843                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 844                                    report, sha1_to_hex(sha1));
 845                 write_or_die(1, buf, len);
 846
 847                 /*
 848                  * Let's just mimic git-unpack-objects here and write
 849                  * the last part of the input buffer to stdout.
 850                  */
 851                 while (input_len) {
 852                         err = xwrite(1, input_buffer + input_offset, input_len);
 853                         if (err <= 0)
 854                                 break;
 855                         input_len -= err;
 856                         input_offset += err;
 857                 }
 858         }
 859 }
 860
 861 static int git_index_pack_config(const char *k, const char *v, void *cb)
 862 {
 863         if (!strcmp(k, "pack.indexversion")) {
 864                 pack_idx_default_version = git_config_int(k, v);
 865                 if (pack_idx_default_version > 2)
 866                         die("bad pack.indexversion=%"PRIu32,
 867                                 pack_idx_default_version);
 868                 return 0;
 869         }
 870         return git_default_config(k, v, cb);
 871 }
 872
 873 int main(int argc, char **argv)
 874 {
 875         int i, fix_thin_pack = 0;
 876         char *curr_pack, *pack_name = NULL;
 877         char *curr_index, *index_name = NULL;
 878         const char *keep_name = NULL, *keep_msg = NULL;
 879         char *index_name_buf = NULL, *keep_name_buf = NULL;
 880         struct pack_idx_entry **idx_objects;
 881         unsigned char pack_sha1[20];
 882
 883         /*
 884          * We wish to read the repository's config file if any, and
 885          * for that it is necessary to call setup_git_directory_gently().
 886          * However if the cwd was inside .git/objects/pack/ then we need
 887          * to go back there or all the pack name arguments will be wrong.
 888          * And in that case we cannot rely on any prefix returned by
 889          * setup_git_directory_gently() either.
 890          */
 891         {
 892                 char cwd[PATH_MAX+1];
 893                 int nongit;
 894
 895                 if (!getcwd(cwd, sizeof(cwd)-1))
 896                         die("Unable to get current working directory");
 897                 setup_git_directory_gently(&nongit);
 898                 git_config(git_index_pack_config, NULL);
 899                 if (chdir(cwd))
 900                         die("Cannot come back to cwd");
 901         }
 902
 903         for (i = 1; i < argc; i++) {
 904                 char *arg = argv[i];
 905
 906                 if (*arg == '-') {
 907                         if (!strcmp(arg, "--stdin")) {
 908                                 from_stdin = 1;
 909                         } else if (!strcmp(arg, "--fix-thin")) {
 910                                 fix_thin_pack = 1;
 911                         } else if (!strcmp(arg, "--strict")) {
 912                                 strict = 1;
 913                         } else if (!strcmp(arg, "--keep")) {
 914                                 keep_msg = "";
 915                         } else if (!prefixcmp(arg, "--keep=")) {
 916                                 keep_msg = arg + 7;
 917                         } else if (!prefixcmp(arg, "--pack_header=")) {
 918                                 struct pack_header *hdr;
 919                                 char *c;
 920
 921                                 hdr = (struct pack_header *)input_buffer;
 922                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 923                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 924                                 if (*c != ',')
 925                                         die("bad %s", arg);
 926                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 927                                 if (*c)
 928                                         die("bad %s", arg);
 929                                 input_len = sizeof(*hdr);
 930                         } else if (!strcmp(arg, "-v")) {
 931                                 verbose = 1;
 932                         } else if (!strcmp(arg, "-o")) {
 933                                 if (index_name || (i+1) >= argc)
 934                                         usage(index_pack_usage);
 935                                 index_name = argv[++i];
 936                         } else if (!prefixcmp(arg, "--index-version=")) {
 937                                 char *c;
 938                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 939                                 if (pack_idx_default_version > 2)
 940                                         die("bad %s", arg);
 941                                 if (*c == ',')
 942                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 943                                 if (*c || pack_idx_off32_limit & 0x80000000)
 944                                         die("bad %s", arg);
 945                         } else
 946                                 usage(index_pack_usage);
 947                         continue;
 948                 }
 949
 950                 if (pack_name)
 951                         usage(index_pack_usage);
 952                 pack_name = arg;
 953         }
 954
 955         if (!pack_name && !from_stdin)
 956                 usage(index_pack_usage);
 957         if (fix_thin_pack && !from_stdin)
 958                 die("--fix-thin cannot be used without --stdin");
 959         if (!index_name && pack_name) {
 960                 int len = strlen(pack_name);
 961                 if (!has_extension(pack_name, ".pack"))
 962                         die("packfile name '%s' does not end with '.pack'",
 963                             pack_name);
 964                 index_name_buf = xmalloc(len);
 965                 memcpy(index_name_buf, pack_name, len - 5);
 966                 strcpy(index_name_buf + len - 5, ".idx");
 967                 index_name = index_name_buf;
 968         }
 969         if (keep_msg && !keep_name && pack_name) {
 970                 int len = strlen(pack_name);
 971                 if (!has_extension(pack_name, ".pack"))
 972                         die("packfile name '%s' does not end with '.pack'",
 973                             pack_name);
 974                 keep_name_buf = xmalloc(len);
 975                 memcpy(keep_name_buf, pack_name, len - 5);
 976                 strcpy(keep_name_buf + len - 5, ".keep");
 977                 keep_name = keep_name_buf;
 978         }
 979
 980         curr_pack = open_pack_file(pack_name);
 981         parse_pack_header();
 982         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 983         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 984         parse_pack_objects(pack_sha1);
 985         if (nr_deltas == nr_resolved_deltas) {
 986                 stop_progress(&progress);
 987                 /* Flush remaining pack final 20-byte SHA1. */
 988                 flush();
 989         } else {
 990                 if (fix_thin_pack) {
 991                         struct sha1file *f;
 992                         unsigned char read_sha1[20], tail_sha1[20];
 993                         char msg[48];
 994                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 995                         int nr_objects_initial = nr_objects;
 996                         if (nr_unresolved <= 0)
 997                                 die("confusion beyond insanity");
 998                         objects = xrealloc(objects,
 999                                            (nr_objects + nr_unresolved + 1)
1000                                            * sizeof(*objects));
1001                         f = sha1fd(output_fd, curr_pack);
1002                         fix_unresolved_deltas(f, nr_unresolved);
1003                         sprintf(msg, "completed with %d local objects",
1004                                 nr_objects - nr_objects_initial);
1005                         stop_progress_msg(&progress, msg);
1006                         sha1close(f, tail_sha1, 0);
1007                         hashcpy(read_sha1, pack_sha1);
1008                         fixup_pack_header_footer(output_fd, pack_sha1,
1009                                                  curr_pack, nr_objects,
1010                                                  read_sha1, consumed_bytes-20);
1011                         if (hashcmp(read_sha1, tail_sha1) != 0)
1012                                 die("Unexpected tail checksum for %s "
1013                                     "(disk corruption?)", curr_pack);
1014                 }
1015                 if (nr_deltas != nr_resolved_deltas)
1016                         die("pack has %d unresolved deltas",
1017                             nr_deltas - nr_resolved_deltas);
1018         }
1019         free(deltas);
1020         if (strict)
1021                 check_objects();
1022
1023         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1024         for (i = 0; i < nr_objects; i++)
1025                 idx_objects[i] = &objects[i].idx;
1026         curr_index = write_idx_file(index_name, idx_objects, nr_objects, pack_sha1);
1027         free(idx_objects);
1028
1029         final(pack_name, curr_pack,
1030                 index_name, curr_index,
1031                 keep_name, keep_msg,
1032                 pack_sha1);
1033         free(objects);
1034         free(index_name_buf);
1035         free(keep_name_buf);
1036         if (pack_name == NULL)
1037                 free(curr_pack);
1038         if (index_name == NULL)
1039                 free(curr_index);
1040
1041         return 0;
1042 }