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
  11 static const char index_pack_usage[] =
  12 "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  13
  14 struct object_entry
  15 {
  16         off_t offset;
  17         unsigned long size;
  18         unsigned int hdr_size;
  19         uint32_t crc32;
  20         enum object_type type;
  21         enum object_type real_type;
  22         unsigned char sha1[20];
  23 };
  24
  25 union delta_base {
  26         unsigned char sha1[20];
  27         off_t offset;
  28 };
  29
  30 /*
  31  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  32  * to memcmp() only the first 20 bytes.
  33  */
  34 #define UNION_BASE_SZ   20
  35
  36 struct delta_entry
  37 {
  38         union delta_base base;
  39         int obj_no;
  40 };
  41
  42 static struct object_entry *objects;
  43 static struct delta_entry *deltas;
  44 static int nr_objects;
  45 static int nr_deltas;
  46 static int nr_resolved_deltas;
  47
  48 static int from_stdin;
  49 static int verbose;
  50
  51 static struct progress progress;
  52
  53 /* We always read in 4kB chunks. */
  54 static unsigned char input_buffer[4096];
  55 static unsigned int input_offset, input_len;
  56 static off_t consumed_bytes;
  57 static SHA_CTX input_ctx;
  58 static uint32_t input_crc32;
  59 static int input_fd, output_fd, pack_fd;
  60
  61 /* Discard current buffer used content. */
  62 static void flush(void)
  63 {
  64         if (input_offset) {
  65                 if (output_fd >= 0)
  66                         write_or_die(output_fd, input_buffer, input_offset);
  67                 SHA1_Update(&input_ctx, input_buffer, input_offset);
  68                 memmove(input_buffer, input_buffer + input_offset, input_len);
  69                 input_offset = 0;
  70         }
  71 }
  72
  73 /*
  74  * Make sure at least "min" bytes are available in the buffer, and
  75  * return the pointer to the buffer.
  76  */
  77 static void *fill(int min)
  78 {
  79         if (min <= input_len)
  80                 return input_buffer + input_offset;
  81         if (min > sizeof(input_buffer))
  82                 die("cannot fill %d bytes", min);
  83         flush();
  84         do {
  85                 ssize_t ret = xread(input_fd, input_buffer + input_len,
  86                                 sizeof(input_buffer) - input_len);
  87                 if (ret <= 0) {
  88                         if (!ret)
  89                                 die("early EOF");
  90                         die("read error on input: %s", strerror(errno));
  91                 }
  92                 input_len += ret;
  93         } while (input_len < min);
  94         return input_buffer;
  95 }
  96
  97 static void use(int bytes)
  98 {
  99         if (bytes > input_len)
 100                 die("used more bytes than were available");
 101         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 102         input_len -= bytes;
 103         input_offset += bytes;
 104
 105         /* make sure off_t is sufficiently large not to wrap */
 106         if (consumed_bytes > consumed_bytes + bytes)
 107                 die("pack too large for current definition of off_t");
 108         consumed_bytes += bytes;
 109 }
 110
 111 static const char *open_pack_file(const char *pack_name)
 112 {
 113         if (from_stdin) {
 114                 input_fd = 0;
 115                 if (!pack_name) {
 116                         static char tmpfile[PATH_MAX];
 117                         snprintf(tmpfile, sizeof(tmpfile),
 118                                  "%s/tmp_pack_XXXXXX", get_object_directory());
 119                         output_fd = mkstemp(tmpfile);
 120                         pack_name = xstrdup(tmpfile);
 121                 } else
 122                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 123                 if (output_fd < 0)
 124                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 125                 pack_fd = output_fd;
 126         } else {
 127                 input_fd = open(pack_name, O_RDONLY);
 128                 if (input_fd < 0)
 129                         die("cannot open packfile '%s': %s",
 130                             pack_name, strerror(errno));
 131                 output_fd = -1;
 132                 pack_fd = input_fd;
 133         }
 134         SHA1_Init(&input_ctx);
 135         return pack_name;
 136 }
 137
 138 static void parse_pack_header(void)
 139 {
 140         struct pack_header *hdr = fill(sizeof(struct pack_header));
 141
 142         /* Header consistency check */
 143         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 144                 die("pack signature mismatch");
 145         if (!pack_version_ok(hdr->hdr_version))
 146                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 147
 148         nr_objects = ntohl(hdr->hdr_entries);
 149         use(sizeof(struct pack_header));
 150 }
 151
 152 static void bad_object(unsigned long offset, const char *format,
 153                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 154
 155 static void bad_object(unsigned long offset, const char *format, ...)
 156 {
 157         va_list params;
 158         char buf[1024];
 159
 160         va_start(params, format);
 161         vsnprintf(buf, sizeof(buf), format, params);
 162         va_end(params);
 163         die("pack has bad object at offset %lu: %s", offset, buf);
 164 }
 165
 166 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 167 {
 168         z_stream stream;
 169         void *buf = xmalloc(size);
 170
 171         memset(&stream, 0, sizeof(stream));
 172         stream.next_out = buf;
 173         stream.avail_out = size;
 174         stream.next_in = fill(1);
 175         stream.avail_in = input_len;
 176         inflateInit(&stream);
 177
 178         for (;;) {
 179                 int ret = inflate(&stream, 0);
 180                 use(input_len - stream.avail_in);
 181                 if (stream.total_out == size && ret == Z_STREAM_END)
 182                         break;
 183                 if (ret != Z_OK)
 184                         bad_object(offset, "inflate returned %d", ret);
 185                 stream.next_in = fill(1);
 186                 stream.avail_in = input_len;
 187         }
 188         inflateEnd(&stream);
 189         return buf;
 190 }
 191
 192 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 193 {
 194         unsigned char *p, c;
 195         unsigned long size;
 196         off_t base_offset;
 197         unsigned shift;
 198         void *data;
 199
 200         obj->offset = consumed_bytes;
 201         input_crc32 = crc32(0, Z_NULL, 0);
 202
 203         p = fill(1);
 204         c = *p;
 205         use(1);
 206         obj->type = (c >> 4) & 7;
 207         size = (c & 15);
 208         shift = 4;
 209         while (c & 0x80) {
 210                 p = fill(1);
 211                 c = *p;
 212                 use(1);
 213                 size += (c & 0x7fUL) << shift;
 214                 shift += 7;
 215         }
 216         obj->size = size;
 217
 218         switch (obj->type) {
 219         case OBJ_REF_DELTA:
 220                 hashcpy(delta_base->sha1, fill(20));
 221                 use(20);
 222                 break;
 223         case OBJ_OFS_DELTA:
 224                 memset(delta_base, 0, sizeof(*delta_base));
 225                 p = fill(1);
 226                 c = *p;
 227                 use(1);
 228                 base_offset = c & 127;
 229                 while (c & 128) {
 230                         base_offset += 1;
 231                         if (!base_offset || MSB(base_offset, 7))
 232                                 bad_object(obj->offset, "offset value overflow for delta base object");
 233                         p = fill(1);
 234                         c = *p;
 235                         use(1);
 236                         base_offset = (base_offset << 7) + (c & 127);
 237                 }
 238                 delta_base->offset = obj->offset - base_offset;
 239                 if (delta_base->offset >= obj->offset)
 240                         bad_object(obj->offset, "delta base offset is out of bound");
 241                 break;
 242         case OBJ_COMMIT:
 243         case OBJ_TREE:
 244         case OBJ_BLOB:
 245         case OBJ_TAG:
 246                 break;
 247         default:
 248                 bad_object(obj->offset, "unknown object type %d", obj->type);
 249         }
 250         obj->hdr_size = consumed_bytes - obj->offset;
 251
 252         data = unpack_entry_data(obj->offset, obj->size);
 253         obj->crc32 = input_crc32;
 254         return data;
 255 }
 256
 257 static void *get_data_from_pack(struct object_entry *obj)
 258 {
 259         unsigned long from = obj[0].offset + obj[0].hdr_size;
 260         unsigned long len = obj[1].offset - from;
 261         unsigned long rdy = 0;
 262         unsigned char *src, *data;
 263         z_stream stream;
 264         int st;
 265
 266         src = xmalloc(len);
 267         data = src;
 268         do {
 269                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 270                 if (n <= 0)
 271                         die("cannot pread pack file: %s", strerror(errno));
 272                 rdy += n;
 273         } while (rdy < len);
 274         data = xmalloc(obj->size);
 275         memset(&stream, 0, sizeof(stream));
 276         stream.next_out = data;
 277         stream.avail_out = obj->size;
 278         stream.next_in = src;
 279         stream.avail_in = len;
 280         inflateInit(&stream);
 281         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 282         inflateEnd(&stream);
 283         if (st != Z_STREAM_END || stream.total_out != obj->size)
 284                 die("serious inflate inconsistency");
 285         free(src);
 286         return data;
 287 }
 288
 289 static int find_delta(const union delta_base *base)
 290 {
 291         int first = 0, last = nr_deltas;
 292
 293         while (first < last) {
 294                 int next = (first + last) / 2;
 295                 struct delta_entry *delta = &deltas[next];
 296                 int cmp;
 297
 298                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 299                 if (!cmp)
 300                         return next;
 301                 if (cmp < 0) {
 302                         last = next;
 303                         continue;
 304                 }
 305                 first = next+1;
 306         }
 307         return -first-1;
 308 }
 309
 310 static int find_delta_children(const union delta_base *base,
 311                                int *first_index, int *last_index)
 312 {
 313         int first = find_delta(base);
 314         int last = first;
 315         int end = nr_deltas - 1;
 316
 317         if (first < 0)
 318                 return -1;
 319         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 320                 --first;
 321         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 322                 ++last;
 323         *first_index = first;
 324         *last_index = last;
 325         return 0;
 326 }
 327
 328 static void sha1_object(const void *data, unsigned long size,
 329                         enum object_type type, unsigned char *sha1)
 330 {
 331         hash_sha1_file(data, size, typename(type), sha1);
 332         if (has_sha1_file(sha1)) {
 333                 void *has_data;
 334                 enum object_type has_type;
 335                 unsigned long has_size;
 336                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 337                 if (!has_data)
 338                         die("cannot read existing object %s", sha1_to_hex(sha1));
 339                 if (size != has_size || type != has_type ||
 340                     memcmp(data, has_data, size) != 0)
 341                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 342                 free(has_data);
 343         }
 344 }
 345
 346 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 347                           unsigned long base_size, enum object_type type)
 348 {
 349         void *delta_data;
 350         unsigned long delta_size;
 351         void *result;
 352         unsigned long result_size;
 353         union delta_base delta_base;
 354         int j, first, last;
 355
 356         delta_obj->real_type = type;
 357         delta_data = get_data_from_pack(delta_obj);
 358         delta_size = delta_obj->size;
 359         result = patch_delta(base_data, base_size, delta_data, delta_size,
 360                              &result_size);
 361         free(delta_data);
 362         if (!result)
 363                 bad_object(delta_obj->offset, "failed to apply delta");
 364         sha1_object(result, result_size, type, delta_obj->sha1);
 365         nr_resolved_deltas++;
 366
 367         hashcpy(delta_base.sha1, delta_obj->sha1);
 368         if (!find_delta_children(&delta_base, &first, &last)) {
 369                 for (j = first; j <= last; j++) {
 370                         struct object_entry *child = objects + deltas[j].obj_no;
 371                         if (child->real_type == OBJ_REF_DELTA)
 372                                 resolve_delta(child, result, result_size, type);
 373                 }
 374         }
 375
 376         memset(&delta_base, 0, sizeof(delta_base));
 377         delta_base.offset = delta_obj->offset;
 378         if (!find_delta_children(&delta_base, &first, &last)) {
 379                 for (j = first; j <= last; j++) {
 380                         struct object_entry *child = objects + deltas[j].obj_no;
 381                         if (child->real_type == OBJ_OFS_DELTA)
 382                                 resolve_delta(child, result, result_size, type);
 383                 }
 384         }
 385
 386         free(result);
 387 }
 388
 389 static int compare_delta_entry(const void *a, const void *b)
 390 {
 391         const struct delta_entry *delta_a = a;
 392         const struct delta_entry *delta_b = b;
 393         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 394 }
 395
 396 /* Parse all objects and return the pack content SHA1 hash */
 397 static void parse_pack_objects(unsigned char *sha1)
 398 {
 399         int i;
 400         struct delta_entry *delta = deltas;
 401         void *data;
 402         struct stat st;
 403
 404         /*
 405          * First pass:
 406          * - find locations of all objects;
 407          * - calculate SHA1 of all non-delta objects;
 408          * - remember base (SHA1 or offset) for all deltas.
 409          */
 410         if (verbose)
 411                 start_progress(&progress, "Indexing %u objects...", "", nr_objects);
 412         for (i = 0; i < nr_objects; i++) {
 413                 struct object_entry *obj = &objects[i];
 414                 data = unpack_raw_entry(obj, &delta->base);
 415                 obj->real_type = obj->type;
 416                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 417                         nr_deltas++;
 418                         delta->obj_no = i;
 419                         delta++;
 420                 } else
 421                         sha1_object(data, obj->size, obj->type, obj->sha1);
 422                 free(data);
 423                 if (verbose)
 424                         display_progress(&progress, i+1);
 425         }
 426         objects[i].offset = consumed_bytes;
 427         if (verbose)
 428                 stop_progress(&progress);
 429
 430         /* Check pack integrity */
 431         flush();
 432         SHA1_Final(sha1, &input_ctx);
 433         if (hashcmp(fill(20), sha1))
 434                 die("pack is corrupted (SHA1 mismatch)");
 435         use(20);
 436
 437         /* If input_fd is a file, we should have reached its end now. */
 438         if (fstat(input_fd, &st))
 439                 die("cannot fstat packfile: %s", strerror(errno));
 440         if (S_ISREG(st.st_mode) &&
 441                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 442                 die("pack has junk at the end");
 443
 444         if (!nr_deltas)
 445                 return;
 446
 447         /* Sort deltas by base SHA1/offset for fast searching */
 448         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 449               compare_delta_entry);
 450
 451         /*
 452          * Second pass:
 453          * - for all non-delta objects, look if it is used as a base for
 454          *   deltas;
 455          * - if used as a base, uncompress the object and apply all deltas,
 456          *   recursively checking if the resulting object is used as a base
 457          *   for some more deltas.
 458          */
 459         if (verbose)
 460                 start_progress(&progress, "Resolving %u deltas...", "", nr_deltas);
 461         for (i = 0; i < nr_objects; i++) {
 462                 struct object_entry *obj = &objects[i];
 463                 union delta_base base;
 464                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 465
 466                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 467                         continue;
 468                 hashcpy(base.sha1, obj->sha1);
 469                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 470                 memset(&base, 0, sizeof(base));
 471                 base.offset = obj->offset;
 472                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 473                 if (!ref && !ofs)
 474                         continue;
 475                 data = get_data_from_pack(obj);
 476                 if (ref)
 477                         for (j = ref_first; j <= ref_last; j++) {
 478                                 struct object_entry *child = objects + deltas[j].obj_no;
 479                                 if (child->real_type == OBJ_REF_DELTA)
 480                                         resolve_delta(child, data,
 481                                                       obj->size, obj->type);
 482                         }
 483                 if (ofs)
 484                         for (j = ofs_first; j <= ofs_last; j++) {
 485                                 struct object_entry *child = objects + deltas[j].obj_no;
 486                                 if (child->real_type == OBJ_OFS_DELTA)
 487                                         resolve_delta(child, data,
 488                                                       obj->size, obj->type);
 489                         }
 490                 free(data);
 491                 if (verbose)
 492                         display_progress(&progress, nr_resolved_deltas);
 493         }
 494 }
 495
 496 static int write_compressed(int fd, void *in, unsigned int size, uint32_t *obj_crc)
 497 {
 498         z_stream stream;
 499         unsigned long maxsize;
 500         void *out;
 501
 502         memset(&stream, 0, sizeof(stream));
 503         deflateInit(&stream, zlib_compression_level);
 504         maxsize = deflateBound(&stream, size);
 505         out = xmalloc(maxsize);
 506
 507         /* Compress it */
 508         stream.next_in = in;
 509         stream.avail_in = size;
 510         stream.next_out = out;
 511         stream.avail_out = maxsize;
 512         while (deflate(&stream, Z_FINISH) == Z_OK);
 513         deflateEnd(&stream);
 514
 515         size = stream.total_out;
 516         write_or_die(fd, out, size);
 517         *obj_crc = crc32(*obj_crc, out, size);
 518         free(out);
 519         return size;
 520 }
 521
 522 static void append_obj_to_pack(const unsigned char *sha1, void *buf,
 523                                unsigned long size, enum object_type type)
 524 {
 525         struct object_entry *obj = &objects[nr_objects++];
 526         unsigned char header[10];
 527         unsigned long s = size;
 528         int n = 0;
 529         unsigned char c = (type << 4) | (s & 15);
 530         s >>= 4;
 531         while (s) {
 532                 header[n++] = c | 0x80;
 533                 c = s & 0x7f;
 534                 s >>= 7;
 535         }
 536         header[n++] = c;
 537         write_or_die(output_fd, header, n);
 538         obj[0].crc32 = crc32(0, Z_NULL, 0);
 539         obj[0].crc32 = crc32(obj[0].crc32, header, n);
 540         obj[1].offset = obj[0].offset + n;
 541         obj[1].offset += write_compressed(output_fd, buf, size, &obj[0].crc32);
 542         hashcpy(obj->sha1, sha1);
 543 }
 544
 545 static int delta_pos_compare(const void *_a, const void *_b)
 546 {
 547         struct delta_entry *a = *(struct delta_entry **)_a;
 548         struct delta_entry *b = *(struct delta_entry **)_b;
 549         return a->obj_no - b->obj_no;
 550 }
 551
 552 static void fix_unresolved_deltas(int nr_unresolved)
 553 {
 554         struct delta_entry **sorted_by_pos;
 555         int i, n = 0;
 556
 557         /*
 558          * Since many unresolved deltas may well be themselves base objects
 559          * for more unresolved deltas, we really want to include the
 560          * smallest number of base objects that would cover as much delta
 561          * as possible by picking the
 562          * trunc deltas first, allowing for other deltas to resolve without
 563          * additional base objects.  Since most base objects are to be found
 564          * before deltas depending on them, a good heuristic is to start
 565          * resolving deltas in the same order as their position in the pack.
 566          */
 567         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 568         for (i = 0; i < nr_deltas; i++) {
 569                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 570                         continue;
 571                 sorted_by_pos[n++] = &deltas[i];
 572         }
 573         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 574
 575         for (i = 0; i < n; i++) {
 576                 struct delta_entry *d = sorted_by_pos[i];
 577                 void *data;
 578                 unsigned long size;
 579                 enum object_type type;
 580                 int j, first, last;
 581
 582                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 583                         continue;
 584                 data = read_sha1_file(d->base.sha1, &type, &size);
 585                 if (!data)
 586                         continue;
 587
 588                 find_delta_children(&d->base, &first, &last);
 589                 for (j = first; j <= last; j++) {
 590                         struct object_entry *child = objects + deltas[j].obj_no;
 591                         if (child->real_type == OBJ_REF_DELTA)
 592                                 resolve_delta(child, data, size, type);
 593                 }
 594
 595                 if (check_sha1_signature(d->base.sha1, data, size, typename(type)))
 596                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 597                 append_obj_to_pack(d->base.sha1, data, size, type);
 598                 free(data);
 599                 if (verbose)
 600                         display_progress(&progress, nr_resolved_deltas);
 601         }
 602         free(sorted_by_pos);
 603 }
 604
 605 static uint32_t index_default_version = 1;
 606 static uint32_t index_off32_limit = 0x7fffffff;
 607
 608 static int sha1_compare(const void *_a, const void *_b)
 609 {
 610         struct object_entry *a = *(struct object_entry **)_a;
 611         struct object_entry *b = *(struct object_entry **)_b;
 612         return hashcmp(a->sha1, b->sha1);
 613 }
 614
 615 /*
 616  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 617  * the SHA1 hash of sorted object names.
 618  */
 619 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 620 {
 621         struct sha1file *f;
 622         struct object_entry **sorted_by_sha, **list, **last;
 623         uint32_t array[256];
 624         int i, fd;
 625         SHA_CTX ctx;
 626         uint32_t index_version;
 627
 628         if (nr_objects) {
 629                 sorted_by_sha =
 630                         xcalloc(nr_objects, sizeof(struct object_entry *));
 631                 list = sorted_by_sha;
 632                 last = sorted_by_sha + nr_objects;
 633                 for (i = 0; i < nr_objects; ++i)
 634                         sorted_by_sha[i] = &objects[i];
 635                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 636                       sha1_compare);
 637         }
 638         else
 639                 sorted_by_sha = list = last = NULL;
 640
 641         if (!index_name) {
 642                 static char tmpfile[PATH_MAX];
 643                 snprintf(tmpfile, sizeof(tmpfile),
 644                          "%s/tmp_idx_XXXXXX", get_object_directory());
 645                 fd = mkstemp(tmpfile);
 646                 index_name = xstrdup(tmpfile);
 647         } else {
 648                 unlink(index_name);
 649                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 650         }
 651         if (fd < 0)
 652                 die("unable to create %s: %s", index_name, strerror(errno));
 653         f = sha1fd(fd, index_name);
 654
 655         /* if last object's offset is >= 2^31 we should use index V2 */
 656         index_version = (objects[nr_objects-1].offset >> 31) ? 2 : index_default_version;
 657
 658         /* index versions 2 and above need a header */
 659         if (index_version >= 2) {
 660                 struct pack_idx_header hdr;
 661                 hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
 662                 hdr.idx_version = htonl(index_version);
 663                 sha1write(f, &hdr, sizeof(hdr));
 664         }
 665
 666         /*
 667          * Write the first-level table (the list is sorted,
 668          * but we use a 256-entry lookup to be able to avoid
 669          * having to do eight extra binary search iterations).
 670          */
 671         for (i = 0; i < 256; i++) {
 672                 struct object_entry **next = list;
 673                 while (next < last) {
 674                         struct object_entry *obj = *next;
 675                         if (obj->sha1[0] != i)
 676                                 break;
 677                         next++;
 678                 }
 679                 array[i] = htonl(next - sorted_by_sha);
 680                 list = next;
 681         }
 682         sha1write(f, array, 256 * 4);
 683
 684         /* compute the SHA1 hash of sorted object names. */
 685         SHA1_Init(&ctx);
 686
 687         /*
 688          * Write the actual SHA1 entries..
 689          */
 690         list = sorted_by_sha;
 691         for (i = 0; i < nr_objects; i++) {
 692                 struct object_entry *obj = *list++;
 693                 if (index_version < 2) {
 694                         uint32_t offset = htonl(obj->offset);
 695                         sha1write(f, &offset, 4);
 696                 }
 697                 sha1write(f, obj->sha1, 20);
 698                 SHA1_Update(&ctx, obj->sha1, 20);
 699         }
 700
 701         if (index_version >= 2) {
 702                 unsigned int nr_large_offset = 0;
 703
 704                 /* write the crc32 table */
 705                 list = sorted_by_sha;
 706                 for (i = 0; i < nr_objects; i++) {
 707                         struct object_entry *obj = *list++;
 708                         uint32_t crc32_val = htonl(obj->crc32);
 709                         sha1write(f, &crc32_val, 4);
 710                 }
 711
 712                 /* write the 32-bit offset table */
 713                 list = sorted_by_sha;
 714                 for (i = 0; i < nr_objects; i++) {
 715                         struct object_entry *obj = *list++;
 716                         uint32_t offset = (obj->offset <= index_off32_limit) ?
 717                                 obj->offset : (0x80000000 | nr_large_offset++);
 718                         offset = htonl(offset);
 719                         sha1write(f, &offset, 4);
 720                 }
 721
 722                 /* write the large offset table */
 723                 list = sorted_by_sha;
 724                 while (nr_large_offset) {
 725                         struct object_entry *obj = *list++;
 726                         uint64_t offset = obj->offset;
 727                         if (offset > index_off32_limit) {
 728                                 uint32_t split[2];
 729                                 split[0]        = htonl(offset >> 32);
 730                                 split[1] = htonl(offset & 0xffffffff);
 731                                 sha1write(f, split, 8);
 732                                 nr_large_offset--;
 733                         }
 734                 }
 735         }
 736
 737         sha1write(f, sha1, 20);
 738         sha1close(f, NULL, 1);
 739         free(sorted_by_sha);
 740         SHA1_Final(sha1, &ctx);
 741         return index_name;
 742 }
 743
 744 static void final(const char *final_pack_name, const char *curr_pack_name,
 745                   const char *final_index_name, const char *curr_index_name,
 746                   const char *keep_name, const char *keep_msg,
 747                   unsigned char *sha1)
 748 {
 749         const char *report = "pack";
 750         char name[PATH_MAX];
 751         int err;
 752
 753         if (!from_stdin) {
 754                 close(input_fd);
 755         } else {
 756                 err = close(output_fd);
 757                 if (err)
 758                         die("error while closing pack file: %s", strerror(errno));
 759                 chmod(curr_pack_name, 0444);
 760         }
 761
 762         if (keep_msg) {
 763                 int keep_fd, keep_msg_len = strlen(keep_msg);
 764                 if (!keep_name) {
 765                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 766                                  get_object_directory(), sha1_to_hex(sha1));
 767                         keep_name = name;
 768                 }
 769                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 770                 if (keep_fd < 0) {
 771                         if (errno != EEXIST)
 772                                 die("cannot write keep file");
 773                 } else {
 774                         if (keep_msg_len > 0) {
 775                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 776                                 write_or_die(keep_fd, "\n", 1);
 777                         }
 778                         close(keep_fd);
 779                         report = "keep";
 780                 }
 781         }
 782
 783         if (final_pack_name != curr_pack_name) {
 784                 if (!final_pack_name) {
 785                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 786                                  get_object_directory(), sha1_to_hex(sha1));
 787                         final_pack_name = name;
 788                 }
 789                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 790                         die("cannot store pack file");
 791         }
 792
 793         chmod(curr_index_name, 0444);
 794         if (final_index_name != curr_index_name) {
 795                 if (!final_index_name) {
 796                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 797                                  get_object_directory(), sha1_to_hex(sha1));
 798                         final_index_name = name;
 799                 }
 800                 if (move_temp_to_file(curr_index_name, final_index_name))
 801                         die("cannot store index file");
 802         }
 803
 804         if (!from_stdin) {
 805                 printf("%s\n", sha1_to_hex(sha1));
 806         } else {
 807                 char buf[48];
 808                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 809                                    report, sha1_to_hex(sha1));
 810                 write_or_die(1, buf, len);
 811
 812                 /*
 813                  * Let's just mimic git-unpack-objects here and write
 814                  * the last part of the input buffer to stdout.
 815                  */
 816                 while (input_len) {
 817                         err = xwrite(1, input_buffer + input_offset, input_len);
 818                         if (err <= 0)
 819                                 break;
 820                         input_len -= err;
 821                         input_offset += err;
 822                 }
 823         }
 824 }
 825
 826 int main(int argc, char **argv)
 827 {
 828         int i, fix_thin_pack = 0;
 829         const char *curr_pack, *pack_name = NULL;
 830         const char *curr_index, *index_name = NULL;
 831         const char *keep_name = NULL, *keep_msg = NULL;
 832         char *index_name_buf = NULL, *keep_name_buf = NULL;
 833         unsigned char sha1[20];
 834
 835         for (i = 1; i < argc; i++) {
 836                 const char *arg = argv[i];
 837
 838                 if (*arg == '-') {
 839                         if (!strcmp(arg, "--stdin")) {
 840                                 from_stdin = 1;
 841                         } else if (!strcmp(arg, "--fix-thin")) {
 842                                 fix_thin_pack = 1;
 843                         } else if (!strcmp(arg, "--keep")) {
 844                                 keep_msg = "";
 845                         } else if (!prefixcmp(arg, "--keep=")) {
 846                                 keep_msg = arg + 7;
 847                         } else if (!prefixcmp(arg, "--pack_header=")) {
 848                                 struct pack_header *hdr;
 849                                 char *c;
 850
 851                                 hdr = (struct pack_header *)input_buffer;
 852                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 853                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 854                                 if (*c != ',')
 855                                         die("bad %s", arg);
 856                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 857                                 if (*c)
 858                                         die("bad %s", arg);
 859                                 input_len = sizeof(*hdr);
 860                         } else if (!strcmp(arg, "-v")) {
 861                                 verbose = 1;
 862                         } else if (!strcmp(arg, "-o")) {
 863                                 if (index_name || (i+1) >= argc)
 864                                         usage(index_pack_usage);
 865                                 index_name = argv[++i];
 866                         } else if (!prefixcmp(arg, "--index-version=")) {
 867                                 char *c;
 868                                 index_default_version = strtoul(arg + 16, &c, 10);
 869                                 if (index_default_version > 2)
 870                                         die("bad %s", arg);
 871                                 if (*c == ',')
 872                                         index_off32_limit = strtoul(c+1, &c, 0);
 873                                 if (*c || index_off32_limit & 0x80000000)
 874                                         die("bad %s", arg);
 875                         } else
 876                                 usage(index_pack_usage);
 877                         continue;
 878                 }
 879
 880                 if (pack_name)
 881                         usage(index_pack_usage);
 882                 pack_name = arg;
 883         }
 884
 885         if (!pack_name && !from_stdin)
 886                 usage(index_pack_usage);
 887         if (fix_thin_pack && !from_stdin)
 888                 die("--fix-thin cannot be used without --stdin");
 889         if (!index_name && pack_name) {
 890                 int len = strlen(pack_name);
 891                 if (!has_extension(pack_name, ".pack"))
 892                         die("packfile name '%s' does not end with '.pack'",
 893                             pack_name);
 894                 index_name_buf = xmalloc(len);
 895                 memcpy(index_name_buf, pack_name, len - 5);
 896                 strcpy(index_name_buf + len - 5, ".idx");
 897                 index_name = index_name_buf;
 898         }
 899         if (keep_msg && !keep_name && pack_name) {
 900                 int len = strlen(pack_name);
 901                 if (!has_extension(pack_name, ".pack"))
 902                         die("packfile name '%s' does not end with '.pack'",
 903                             pack_name);
 904                 keep_name_buf = xmalloc(len);
 905                 memcpy(keep_name_buf, pack_name, len - 5);
 906                 strcpy(keep_name_buf + len - 5, ".keep");
 907                 keep_name = keep_name_buf;
 908         }
 909
 910         curr_pack = open_pack_file(pack_name);
 911         parse_pack_header();
 912         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 913         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 914         parse_pack_objects(sha1);
 915         if (nr_deltas == nr_resolved_deltas) {
 916                 if (verbose)
 917                         stop_progress(&progress);
 918                 /* Flush remaining pack final 20-byte SHA1. */
 919                 flush();
 920         } else {
 921                 if (fix_thin_pack) {
 922                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 923                         int nr_objects_initial = nr_objects;
 924                         if (nr_unresolved <= 0)
 925                                 die("confusion beyond insanity");
 926                         objects = xrealloc(objects,
 927                                            (nr_objects + nr_unresolved + 1)
 928                                            * sizeof(*objects));
 929                         fix_unresolved_deltas(nr_unresolved);
 930                         if (verbose) {
 931                                 stop_progress(&progress);
 932                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 933                                         nr_objects - nr_objects_initial);
 934                         }
 935                         fixup_pack_header_footer(output_fd, sha1,
 936                                 curr_pack, nr_objects);
 937                 }
 938                 if (nr_deltas != nr_resolved_deltas)
 939                         die("pack has %d unresolved deltas",
 940                             nr_deltas - nr_resolved_deltas);
 941         }
 942         free(deltas);
 943         curr_index = write_index_file(index_name, sha1);
 944         final(pack_name, curr_pack,
 945                 index_name, curr_index,
 946                 keep_name, keep_msg,
 947                 sha1);
 948         free(objects);
 949         free(index_name_buf);
 950         free(keep_name_buf);
 951
 952         return 0;
 953 }