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 /*
  30  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  31  * to memcmp() only the first 20 bytes.
  32  */
  33 #define UNION_BASE_SZ   20
  34
  35 #define FLAG_LINK (1u<<20)
  36 #define FLAG_CHECKED (1u<<21)
  37
  38 struct delta_entry
  39 {
  40         union delta_base base;
  41         int obj_no;
  42 };
  43
  44 static struct object_entry *objects;
  45 static struct delta_entry *deltas;
  46 static int nr_objects;
  47 static int nr_deltas;
  48 static int nr_resolved_deltas;
  49
  50 static int from_stdin;
  51 static int strict;
  52 static int verbose;
  53
  54 static struct progress *progress;
  55
  56 /* We always read in 4kB chunks. */
  57 static unsigned char input_buffer[4096];
  58 static unsigned int input_offset, input_len;
  59 static off_t consumed_bytes;
  60 static SHA_CTX input_ctx;
  61 static uint32_t input_crc32;
  62 static int input_fd, output_fd, pack_fd;
  63
  64 static int mark_link(struct object *obj, int type, void *data)
  65 {
  66         if (!obj)
  67                 return -1;
  68
  69         if (type != OBJ_ANY && obj->type != type)
  70                 die("object type mismatch at %s", sha1_to_hex(obj->sha1));
  71
  72         obj->flags |= FLAG_LINK;
  73         return 0;
  74 }
  75
  76 /* The content of each linked object must have been checked
  77    or it must be already present in the object database */
  78 static void check_object(struct object *obj)
  79 {
  80         if (!obj)
  81                 return;
  82
  83         if (!(obj->flags & FLAG_LINK))
  84                 return;
  85
  86         if (!(obj->flags & FLAG_CHECKED)) {
  87                 unsigned long size;
  88                 int type = sha1_object_info(obj->sha1, &size);
  89                 if (type != obj->type || type <= 0)
  90                         die("object of unexpected type");
  91                 obj->flags |= FLAG_CHECKED;
  92                 return;
  93         }
  94 }
  95
  96 static void check_objects(void)
  97 {
  98         unsigned i, max;
  99
 100         max = get_max_object_index();
 101         for (i = 0; i < max; i++)
 102                 check_object(get_indexed_object(i));
 103 }
 104
 105
 106 /* Discard current buffer used content. */
 107 static void flush(void)
 108 {
 109         if (input_offset) {
 110                 if (output_fd >= 0)
 111                         write_or_die(output_fd, input_buffer, input_offset);
 112                 SHA1_Update(&input_ctx, input_buffer, input_offset);
 113                 memmove(input_buffer, input_buffer + input_offset, input_len);
 114                 input_offset = 0;
 115         }
 116 }
 117
 118 /*
 119  * Make sure at least "min" bytes are available in the buffer, and
 120  * return the pointer to the buffer.
 121  */
 122 static void *fill(int min)
 123 {
 124         if (min <= input_len)
 125                 return input_buffer + input_offset;
 126         if (min > sizeof(input_buffer))
 127                 die("cannot fill %d bytes", min);
 128         flush();
 129         do {
 130                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 131                                 sizeof(input_buffer) - input_len);
 132                 if (ret <= 0) {
 133                         if (!ret)
 134                                 die("early EOF");
 135                         die("read error on input: %s", strerror(errno));
 136                 }
 137                 input_len += ret;
 138                 if (from_stdin)
 139                         display_throughput(progress, consumed_bytes + input_len);
 140         } while (input_len < min);
 141         return input_buffer;
 142 }
 143
 144 static void use(int bytes)
 145 {
 146         if (bytes > input_len)
 147                 die("used more bytes than were available");
 148         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 149         input_len -= bytes;
 150         input_offset += bytes;
 151
 152         /* make sure off_t is sufficiently large not to wrap */
 153         if (consumed_bytes > consumed_bytes + bytes)
 154                 die("pack too large for current definition of off_t");
 155         consumed_bytes += bytes;
 156 }
 157
 158 static char *open_pack_file(char *pack_name)
 159 {
 160         if (from_stdin) {
 161                 input_fd = 0;
 162                 if (!pack_name) {
 163                         static char tmpfile[PATH_MAX];
 164                         snprintf(tmpfile, sizeof(tmpfile),
 165                                  "%s/tmp_pack_XXXXXX", get_object_directory());
 166                         output_fd = xmkstemp(tmpfile);
 167                         pack_name = xstrdup(tmpfile);
 168                 } else
 169                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 170                 if (output_fd < 0)
 171                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 172                 pack_fd = output_fd;
 173         } else {
 174                 input_fd = open(pack_name, O_RDONLY);
 175                 if (input_fd < 0)
 176                         die("cannot open packfile '%s': %s",
 177                             pack_name, strerror(errno));
 178                 output_fd = -1;
 179                 pack_fd = input_fd;
 180         }
 181         SHA1_Init(&input_ctx);
 182         return pack_name;
 183 }
 184
 185 static void parse_pack_header(void)
 186 {
 187         struct pack_header *hdr = fill(sizeof(struct pack_header));
 188
 189         /* Header consistency check */
 190         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 191                 die("pack signature mismatch");
 192         if (!pack_version_ok(hdr->hdr_version))
 193                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 194
 195         nr_objects = ntohl(hdr->hdr_entries);
 196         use(sizeof(struct pack_header));
 197 }
 198
 199 static void bad_object(unsigned long offset, const char *format,
 200                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 201
 202 static void bad_object(unsigned long offset, const char *format, ...)
 203 {
 204         va_list params;
 205         char buf[1024];
 206
 207         va_start(params, format);
 208         vsnprintf(buf, sizeof(buf), format, params);
 209         va_end(params);
 210         die("pack has bad object at offset %lu: %s", offset, buf);
 211 }
 212
 213 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 214 {
 215         z_stream stream;
 216         void *buf = xmalloc(size);
 217
 218         memset(&stream, 0, sizeof(stream));
 219         stream.next_out = buf;
 220         stream.avail_out = size;
 221         stream.next_in = fill(1);
 222         stream.avail_in = input_len;
 223         inflateInit(&stream);
 224
 225         for (;;) {
 226                 int ret = inflate(&stream, 0);
 227                 use(input_len - stream.avail_in);
 228                 if (stream.total_out == size && ret == Z_STREAM_END)
 229                         break;
 230                 if (ret != Z_OK)
 231                         bad_object(offset, "inflate returned %d", ret);
 232                 stream.next_in = fill(1);
 233                 stream.avail_in = input_len;
 234         }
 235         inflateEnd(&stream);
 236         return buf;
 237 }
 238
 239 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 240 {
 241         unsigned char *p, c;
 242         unsigned long size;
 243         off_t base_offset;
 244         unsigned shift;
 245         void *data;
 246
 247         obj->idx.offset = consumed_bytes;
 248         input_crc32 = crc32(0, Z_NULL, 0);
 249
 250         p = fill(1);
 251         c = *p;
 252         use(1);
 253         obj->type = (c >> 4) & 7;
 254         size = (c & 15);
 255         shift = 4;
 256         while (c & 0x80) {
 257                 p = fill(1);
 258                 c = *p;
 259                 use(1);
 260                 size += (c & 0x7fUL) << shift;
 261                 shift += 7;
 262         }
 263         obj->size = size;
 264
 265         switch (obj->type) {
 266         case OBJ_REF_DELTA:
 267                 hashcpy(delta_base->sha1, fill(20));
 268                 use(20);
 269                 break;
 270         case OBJ_OFS_DELTA:
 271                 memset(delta_base, 0, sizeof(*delta_base));
 272                 p = fill(1);
 273                 c = *p;
 274                 use(1);
 275                 base_offset = c & 127;
 276                 while (c & 128) {
 277                         base_offset += 1;
 278                         if (!base_offset || MSB(base_offset, 7))
 279                                 bad_object(obj->idx.offset, "offset value overflow for delta base object");
 280                         p = fill(1);
 281                         c = *p;
 282                         use(1);
 283                         base_offset = (base_offset << 7) + (c & 127);
 284                 }
 285                 delta_base->offset = obj->idx.offset - base_offset;
 286                 if (delta_base->offset >= obj->idx.offset)
 287                         bad_object(obj->idx.offset, "delta base offset is out of bound");
 288                 break;
 289         case OBJ_COMMIT:
 290         case OBJ_TREE:
 291         case OBJ_BLOB:
 292         case OBJ_TAG:
 293                 break;
 294         default:
 295                 bad_object(obj->idx.offset, "unknown object type %d", obj->type);
 296         }
 297         obj->hdr_size = consumed_bytes - obj->idx.offset;
 298
 299         data = unpack_entry_data(obj->idx.offset, obj->size);
 300         obj->idx.crc32 = input_crc32;
 301         return data;
 302 }
 303
 304 static void *get_data_from_pack(struct object_entry *obj)
 305 {
 306         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 307         unsigned long len = obj[1].idx.offset - from;
 308         unsigned long rdy = 0;
 309         unsigned char *src, *data;
 310         z_stream stream;
 311         int st;
 312
 313         src = xmalloc(len);
 314         data = src;
 315         do {
 316                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 317                 if (n <= 0)
 318                         die("cannot pread pack file: %s", strerror(errno));
 319                 rdy += n;
 320         } while (rdy < len);
 321         data = xmalloc(obj->size);
 322         memset(&stream, 0, sizeof(stream));
 323         stream.next_out = data;
 324         stream.avail_out = obj->size;
 325         stream.next_in = src;
 326         stream.avail_in = len;
 327         inflateInit(&stream);
 328         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 329         inflateEnd(&stream);
 330         if (st != Z_STREAM_END || stream.total_out != obj->size)
 331                 die("serious inflate inconsistency");
 332         free(src);
 333         return data;
 334 }
 335
 336 static int find_delta(const union delta_base *base)
 337 {
 338         int first = 0, last = nr_deltas;
 339
 340         while (first < last) {
 341                 int next = (first + last) / 2;
 342                 struct delta_entry *delta = &deltas[next];
 343                 int cmp;
 344
 345                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 346                 if (!cmp)
 347                         return next;
 348                 if (cmp < 0) {
 349                         last = next;
 350                         continue;
 351                 }
 352                 first = next+1;
 353         }
 354         return -first-1;
 355 }
 356
 357 static int find_delta_children(const union delta_base *base,
 358                                int *first_index, int *last_index)
 359 {
 360         int first = find_delta(base);
 361         int last = first;
 362         int end = nr_deltas - 1;
 363
 364         if (first < 0)
 365                 return -1;
 366         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 367                 --first;
 368         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 369                 ++last;
 370         *first_index = first;
 371         *last_index = last;
 372         return 0;
 373 }
 374
 375 static void sha1_object(const void *data, unsigned long size,
 376                         enum object_type type, unsigned char *sha1)
 377 {
 378         hash_sha1_file(data, size, typename(type), sha1);
 379         if (has_sha1_file(sha1)) {
 380                 void *has_data;
 381                 enum object_type has_type;
 382                 unsigned long has_size;
 383                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 384                 if (!has_data)
 385                         die("cannot read existing object %s", sha1_to_hex(sha1));
 386                 if (size != has_size || type != has_type ||
 387                     memcmp(data, has_data, size) != 0)
 388                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 389                 free(has_data);
 390         }
 391         if (strict) {
 392                 if (type == OBJ_BLOB) {
 393                         struct blob *blob = lookup_blob(sha1);
 394                         if (blob)
 395                                 blob->object.flags |= FLAG_CHECKED;
 396                         else
 397                                 die("invalid blob object %s", sha1_to_hex(sha1));
 398                 } else {
 399                         struct object *obj;
 400                         int eaten;
 401                         void *buf = (void *) data;
 402
 403                         /*
 404                          * we do not need to free the memory here, as the
 405                          * buf is deleted by the caller.
 406                          */
 407                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 408                         if (!obj)
 409                                 die("invalid %s", typename(type));
 410                         if (fsck_object(obj, 1, fsck_error_function))
 411                                 die("Error in object");
 412                         if (fsck_walk(obj, mark_link, 0))
 413                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 414
 415                         if (obj->type == OBJ_TREE) {
 416                                 struct tree *item = (struct tree *) obj;
 417                                 item->buffer = NULL;
 418                         }
 419                         if (obj->type == OBJ_COMMIT) {
 420                                 struct commit *commit = (struct commit *) obj;
 421                                 commit->buffer = NULL;
 422                         }
 423                         obj->flags |= FLAG_CHECKED;
 424                 }
 425         }
 426 }
 427
 428 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 429                           unsigned long base_size, enum object_type type)
 430 {
 431         void *delta_data;
 432         unsigned long delta_size;
 433         void *result;
 434         unsigned long result_size;
 435         union delta_base delta_base;
 436         int j, first, last;
 437
 438         delta_obj->real_type = type;
 439         delta_data = get_data_from_pack(delta_obj);
 440         delta_size = delta_obj->size;
 441         result = patch_delta(base_data, base_size, delta_data, delta_size,
 442                              &result_size);
 443         free(delta_data);
 444         if (!result)
 445                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 446         sha1_object(result, result_size, type, delta_obj->idx.sha1);
 447         nr_resolved_deltas++;
 448
 449         hashcpy(delta_base.sha1, delta_obj->idx.sha1);
 450         if (!find_delta_children(&delta_base, &first, &last)) {
 451                 for (j = first; j <= last; j++) {
 452                         struct object_entry *child = objects + deltas[j].obj_no;
 453                         if (child->real_type == OBJ_REF_DELTA)
 454                                 resolve_delta(child, result, result_size, type);
 455                 }
 456         }
 457
 458         memset(&delta_base, 0, sizeof(delta_base));
 459         delta_base.offset = delta_obj->idx.offset;
 460         if (!find_delta_children(&delta_base, &first, &last)) {
 461                 for (j = first; j <= last; j++) {
 462                         struct object_entry *child = objects + deltas[j].obj_no;
 463                         if (child->real_type == OBJ_OFS_DELTA)
 464                                 resolve_delta(child, result, result_size, type);
 465                 }
 466         }
 467
 468         free(result);
 469 }
 470
 471 static int compare_delta_entry(const void *a, const void *b)
 472 {
 473         const struct delta_entry *delta_a = a;
 474         const struct delta_entry *delta_b = b;
 475         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 476 }
 477
 478 /* Parse all objects and return the pack content SHA1 hash */
 479 static void parse_pack_objects(unsigned char *sha1)
 480 {
 481         int i;
 482         struct delta_entry *delta = deltas;
 483         void *data;
 484         struct stat st;
 485
 486         /*
 487          * First pass:
 488          * - find locations of all objects;
 489          * - calculate SHA1 of all non-delta objects;
 490          * - remember base (SHA1 or offset) for all deltas.
 491          */
 492         if (verbose)
 493                 progress = start_progress(
 494                                 from_stdin ? "Receiving objects" : "Indexing objects",
 495                                 nr_objects);
 496         for (i = 0; i < nr_objects; i++) {
 497                 struct object_entry *obj = &objects[i];
 498                 data = unpack_raw_entry(obj, &delta->base);
 499                 obj->real_type = obj->type;
 500                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 501                         nr_deltas++;
 502                         delta->obj_no = i;
 503                         delta++;
 504                 } else
 505                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 506                 free(data);
 507                 display_progress(progress, i+1);
 508         }
 509         objects[i].idx.offset = consumed_bytes;
 510         stop_progress(&progress);
 511
 512         /* Check pack integrity */
 513         flush();
 514         SHA1_Final(sha1, &input_ctx);
 515         if (hashcmp(fill(20), sha1))
 516                 die("pack is corrupted (SHA1 mismatch)");
 517         use(20);
 518
 519         /* If input_fd is a file, we should have reached its end now. */
 520         if (fstat(input_fd, &st))
 521                 die("cannot fstat packfile: %s", strerror(errno));
 522         if (S_ISREG(st.st_mode) &&
 523                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 524                 die("pack has junk at the end");
 525
 526         if (!nr_deltas)
 527                 return;
 528
 529         /* Sort deltas by base SHA1/offset for fast searching */
 530         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 531               compare_delta_entry);
 532
 533         /*
 534          * Second pass:
 535          * - for all non-delta objects, look if it is used as a base for
 536          *   deltas;
 537          * - if used as a base, uncompress the object and apply all deltas,
 538          *   recursively checking if the resulting object is used as a base
 539          *   for some more deltas.
 540          */
 541         if (verbose)
 542                 progress = start_progress("Resolving deltas", nr_deltas);
 543         for (i = 0; i < nr_objects; i++) {
 544                 struct object_entry *obj = &objects[i];
 545                 union delta_base base;
 546                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 547
 548                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 549                         continue;
 550                 hashcpy(base.sha1, obj->idx.sha1);
 551                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 552                 memset(&base, 0, sizeof(base));
 553                 base.offset = obj->idx.offset;
 554                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 555                 if (!ref && !ofs)
 556                         continue;
 557                 data = get_data_from_pack(obj);
 558                 if (ref)
 559                         for (j = ref_first; j <= ref_last; j++) {
 560                                 struct object_entry *child = objects + deltas[j].obj_no;
 561                                 if (child->real_type == OBJ_REF_DELTA)
 562                                         resolve_delta(child, data,
 563                                                       obj->size, obj->type);
 564                         }
 565                 if (ofs)
 566                         for (j = ofs_first; j <= ofs_last; j++) {
 567                                 struct object_entry *child = objects + deltas[j].obj_no;
 568                                 if (child->real_type == OBJ_OFS_DELTA)
 569                                         resolve_delta(child, data,
 570                                                       obj->size, obj->type);
 571                         }
 572                 free(data);
 573                 display_progress(progress, nr_resolved_deltas);
 574         }
 575 }
 576
 577 static int write_compressed(int fd, void *in, unsigned int size, uint32_t *obj_crc)
 578 {
 579         z_stream stream;
 580         unsigned long maxsize;
 581         void *out;
 582
 583         memset(&stream, 0, sizeof(stream));
 584         deflateInit(&stream, zlib_compression_level);
 585         maxsize = deflateBound(&stream, size);
 586         out = xmalloc(maxsize);
 587
 588         /* Compress it */
 589         stream.next_in = in;
 590         stream.avail_in = size;
 591         stream.next_out = out;
 592         stream.avail_out = maxsize;
 593         while (deflate(&stream, Z_FINISH) == Z_OK);
 594         deflateEnd(&stream);
 595
 596         size = stream.total_out;
 597         write_or_die(fd, out, size);
 598         *obj_crc = crc32(*obj_crc, out, size);
 599         free(out);
 600         return size;
 601 }
 602
 603 static void append_obj_to_pack(const unsigned char *sha1, void *buf,
 604                                unsigned long size, enum object_type type)
 605 {
 606         struct object_entry *obj = &objects[nr_objects++];
 607         unsigned char header[10];
 608         unsigned long s = size;
 609         int n = 0;
 610         unsigned char c = (type << 4) | (s & 15);
 611         s >>= 4;
 612         while (s) {
 613                 header[n++] = c | 0x80;
 614                 c = s & 0x7f;
 615                 s >>= 7;
 616         }
 617         header[n++] = c;
 618         write_or_die(output_fd, header, n);
 619         obj[0].idx.crc32 = crc32(0, Z_NULL, 0);
 620         obj[0].idx.crc32 = crc32(obj[0].idx.crc32, header, n);
 621         obj[1].idx.offset = obj[0].idx.offset + n;
 622         obj[1].idx.offset += write_compressed(output_fd, buf, size, &obj[0].idx.crc32);
 623         hashcpy(obj->idx.sha1, sha1);
 624 }
 625
 626 static int delta_pos_compare(const void *_a, const void *_b)
 627 {
 628         struct delta_entry *a = *(struct delta_entry **)_a;
 629         struct delta_entry *b = *(struct delta_entry **)_b;
 630         return a->obj_no - b->obj_no;
 631 }
 632
 633 static void fix_unresolved_deltas(int nr_unresolved)
 634 {
 635         struct delta_entry **sorted_by_pos;
 636         int i, n = 0;
 637
 638         /*
 639          * Since many unresolved deltas may well be themselves base objects
 640          * for more unresolved deltas, we really want to include the
 641          * smallest number of base objects that would cover as much delta
 642          * as possible by picking the
 643          * trunc deltas first, allowing for other deltas to resolve without
 644          * additional base objects.  Since most base objects are to be found
 645          * before deltas depending on them, a good heuristic is to start
 646          * resolving deltas in the same order as their position in the pack.
 647          */
 648         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 649         for (i = 0; i < nr_deltas; i++) {
 650                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 651                         continue;
 652                 sorted_by_pos[n++] = &deltas[i];
 653         }
 654         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 655
 656         for (i = 0; i < n; i++) {
 657                 struct delta_entry *d = sorted_by_pos[i];
 658                 void *data;
 659                 unsigned long size;
 660                 enum object_type type;
 661                 int j, first, last;
 662
 663                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 664                         continue;
 665                 data = read_sha1_file(d->base.sha1, &type, &size);
 666                 if (!data)
 667                         continue;
 668
 669                 find_delta_children(&d->base, &first, &last);
 670                 for (j = first; j <= last; j++) {
 671                         struct object_entry *child = objects + deltas[j].obj_no;
 672                         if (child->real_type == OBJ_REF_DELTA)
 673                                 resolve_delta(child, data, size, type);
 674                 }
 675
 676                 if (check_sha1_signature(d->base.sha1, data, size, typename(type)))
 677                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 678                 append_obj_to_pack(d->base.sha1, data, size, type);
 679                 free(data);
 680                 display_progress(progress, nr_resolved_deltas);
 681         }
 682         free(sorted_by_pos);
 683 }
 684
 685 static void final(const char *final_pack_name, const char *curr_pack_name,
 686                   const char *final_index_name, const char *curr_index_name,
 687                   const char *keep_name, const char *keep_msg,
 688                   unsigned char *sha1)
 689 {
 690         const char *report = "pack";
 691         char name[PATH_MAX];
 692         int err;
 693
 694         if (!from_stdin) {
 695                 close(input_fd);
 696         } else {
 697                 fsync_or_die(output_fd, curr_pack_name);
 698                 err = close(output_fd);
 699                 if (err)
 700                         die("error while closing pack file: %s", strerror(errno));
 701                 chmod(curr_pack_name, 0444);
 702         }
 703
 704         if (keep_msg) {
 705                 int keep_fd, keep_msg_len = strlen(keep_msg);
 706                 if (!keep_name) {
 707                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 708                                  get_object_directory(), sha1_to_hex(sha1));
 709                         keep_name = name;
 710                 }
 711                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 712                 if (keep_fd < 0) {
 713                         if (errno != EEXIST)
 714                                 die("cannot write keep file");
 715                 } else {
 716                         if (keep_msg_len > 0) {
 717                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 718                                 write_or_die(keep_fd, "\n", 1);
 719                         }
 720                         if (close(keep_fd) != 0)
 721                                 die("cannot write keep file");
 722                         report = "keep";
 723                 }
 724         }
 725
 726         if (final_pack_name != curr_pack_name) {
 727                 if (!final_pack_name) {
 728                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 729                                  get_object_directory(), sha1_to_hex(sha1));
 730                         final_pack_name = name;
 731                 }
 732                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 733                         die("cannot store pack file");
 734         }
 735
 736         chmod(curr_index_name, 0444);
 737         if (final_index_name != curr_index_name) {
 738                 if (!final_index_name) {
 739                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 740                                  get_object_directory(), sha1_to_hex(sha1));
 741                         final_index_name = name;
 742                 }
 743                 if (move_temp_to_file(curr_index_name, final_index_name))
 744                         die("cannot store index file");
 745         }
 746
 747         if (!from_stdin) {
 748                 printf("%s\n", sha1_to_hex(sha1));
 749         } else {
 750                 char buf[48];
 751                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 752                                    report, sha1_to_hex(sha1));
 753                 write_or_die(1, buf, len);
 754
 755                 /*
 756                  * Let's just mimic git-unpack-objects here and write
 757                  * the last part of the input buffer to stdout.
 758                  */
 759                 while (input_len) {
 760                         err = xwrite(1, input_buffer + input_offset, input_len);
 761                         if (err <= 0)
 762                                 break;
 763                         input_len -= err;
 764                         input_offset += err;
 765                 }
 766         }
 767 }
 768
 769 static int git_index_pack_config(const char *k, const char *v, void *cb)
 770 {
 771         if (!strcmp(k, "pack.indexversion")) {
 772                 pack_idx_default_version = git_config_int(k, v);
 773                 if (pack_idx_default_version > 2)
 774                         die("bad pack.indexversion=%d", pack_idx_default_version);
 775                 return 0;
 776         }
 777         return git_default_config(k, v, cb);
 778 }
 779
 780 int main(int argc, char **argv)
 781 {
 782         int i, fix_thin_pack = 0;
 783         char *curr_pack, *pack_name = NULL;
 784         char *curr_index, *index_name = NULL;
 785         const char *keep_name = NULL, *keep_msg = NULL;
 786         char *index_name_buf = NULL, *keep_name_buf = NULL;
 787         struct pack_idx_entry **idx_objects;
 788         unsigned char sha1[20];
 789
 790         git_config(git_index_pack_config, NULL);
 791
 792         for (i = 1; i < argc; i++) {
 793                 char *arg = argv[i];
 794
 795                 if (*arg == '-') {
 796                         if (!strcmp(arg, "--stdin")) {
 797                                 from_stdin = 1;
 798                         } else if (!strcmp(arg, "--fix-thin")) {
 799                                 fix_thin_pack = 1;
 800                         } else if (!strcmp(arg, "--strict")) {
 801                                 strict = 1;
 802                         } else if (!strcmp(arg, "--keep")) {
 803                                 keep_msg = "";
 804                         } else if (!prefixcmp(arg, "--keep=")) {
 805                                 keep_msg = arg + 7;
 806                         } else if (!prefixcmp(arg, "--pack_header=")) {
 807                                 struct pack_header *hdr;
 808                                 char *c;
 809
 810                                 hdr = (struct pack_header *)input_buffer;
 811                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 812                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 813                                 if (*c != ',')
 814                                         die("bad %s", arg);
 815                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 816                                 if (*c)
 817                                         die("bad %s", arg);
 818                                 input_len = sizeof(*hdr);
 819                         } else if (!strcmp(arg, "-v")) {
 820                                 verbose = 1;
 821                         } else if (!strcmp(arg, "-o")) {
 822                                 if (index_name || (i+1) >= argc)
 823                                         usage(index_pack_usage);
 824                                 index_name = argv[++i];
 825                         } else if (!prefixcmp(arg, "--index-version=")) {
 826                                 char *c;
 827                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 828                                 if (pack_idx_default_version > 2)
 829                                         die("bad %s", arg);
 830                                 if (*c == ',')
 831                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 832                                 if (*c || pack_idx_off32_limit & 0x80000000)
 833                                         die("bad %s", arg);
 834                         } else
 835                                 usage(index_pack_usage);
 836                         continue;
 837                 }
 838
 839                 if (pack_name)
 840                         usage(index_pack_usage);
 841                 pack_name = arg;
 842         }
 843
 844         if (!pack_name && !from_stdin)
 845                 usage(index_pack_usage);
 846         if (fix_thin_pack && !from_stdin)
 847                 die("--fix-thin cannot be used without --stdin");
 848         if (!index_name && pack_name) {
 849                 int len = strlen(pack_name);
 850                 if (!has_extension(pack_name, ".pack"))
 851                         die("packfile name '%s' does not end with '.pack'",
 852                             pack_name);
 853                 index_name_buf = xmalloc(len);
 854                 memcpy(index_name_buf, pack_name, len - 5);
 855                 strcpy(index_name_buf + len - 5, ".idx");
 856                 index_name = index_name_buf;
 857         }
 858         if (keep_msg && !keep_name && pack_name) {
 859                 int len = strlen(pack_name);
 860                 if (!has_extension(pack_name, ".pack"))
 861                         die("packfile name '%s' does not end with '.pack'",
 862                             pack_name);
 863                 keep_name_buf = xmalloc(len);
 864                 memcpy(keep_name_buf, pack_name, len - 5);
 865                 strcpy(keep_name_buf + len - 5, ".keep");
 866                 keep_name = keep_name_buf;
 867         }
 868
 869         curr_pack = open_pack_file(pack_name);
 870         parse_pack_header();
 871         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 872         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 873         parse_pack_objects(sha1);
 874         if (nr_deltas == nr_resolved_deltas) {
 875                 stop_progress(&progress);
 876                 /* Flush remaining pack final 20-byte SHA1. */
 877                 flush();
 878         } else {
 879                 if (fix_thin_pack) {
 880                         char msg[48];
 881                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 882                         int nr_objects_initial = nr_objects;
 883                         if (nr_unresolved <= 0)
 884                                 die("confusion beyond insanity");
 885                         objects = xrealloc(objects,
 886                                            (nr_objects + nr_unresolved + 1)
 887                                            * sizeof(*objects));
 888                         fix_unresolved_deltas(nr_unresolved);
 889                         sprintf(msg, "completed with %d local objects",
 890                                 nr_objects - nr_objects_initial);
 891                         stop_progress_msg(&progress, msg);
 892                         fixup_pack_header_footer(output_fd, sha1,
 893                                                  curr_pack, nr_objects);
 894                 }
 895                 if (nr_deltas != nr_resolved_deltas)
 896                         die("pack has %d unresolved deltas",
 897                             nr_deltas - nr_resolved_deltas);
 898         }
 899         free(deltas);
 900         if (strict)
 901                 check_objects();
 902
 903         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
 904         for (i = 0; i < nr_objects; i++)
 905                 idx_objects[i] = &objects[i].idx;
 906         curr_index = write_idx_file(index_name, idx_objects, nr_objects, sha1);
 907         free(idx_objects);
 908
 909         final(pack_name, curr_pack,
 910                 index_name, curr_index,
 911                 keep_name, keep_msg,
 912                 sha1);
 913         free(objects);
 914         free(index_name_buf);
 915         free(keep_name_buf);
 916         if (pack_name == NULL)
 917                 free(curr_pack);
 918         if (index_name == NULL)
 919                 free(curr_index);
 920
 921         return 0;
 922 }