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
  10 static const char index_pack_usage[] =
  11 "git-index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  12
  13 struct object_entry
  14 {
  15         off_t offset;
  16         unsigned long size;
  17         unsigned int hdr_size;
  18         uint32_t crc32;
  19         enum object_type type;
  20         enum object_type real_type;
  21         unsigned char sha1[20];
  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 struct delta_entry
  36 {
  37         union delta_base base;
  38         int obj_no;
  39 };
  40
  41 static struct object_entry *objects;
  42 static struct delta_entry *deltas;
  43 static int nr_objects;
  44 static int nr_deltas;
  45 static int nr_resolved_deltas;
  46
  47 static int from_stdin;
  48 static int verbose;
  49
  50 static volatile sig_atomic_t progress_update;
  51
  52 static void progress_interval(int signum)
  53 {
  54         progress_update = 1;
  55 }
  56
  57 static void setup_progress_signal(void)
  58 {
  59         struct sigaction sa;
  60         struct itimerval v;
  61
  62         memset(&sa, 0, sizeof(sa));
  63         sa.sa_handler = progress_interval;
  64         sigemptyset(&sa.sa_mask);
  65         sa.sa_flags = SA_RESTART;
  66         sigaction(SIGALRM, &sa, NULL);
  67
  68         v.it_interval.tv_sec = 1;
  69         v.it_interval.tv_usec = 0;
  70         v.it_value = v.it_interval;
  71         setitimer(ITIMER_REAL, &v, NULL);
  72
  73 }
  74
  75 static unsigned display_progress(unsigned n, unsigned total, unsigned last_pc)
  76 {
  77         unsigned percent = n * 100 / total;
  78         if (percent != last_pc || progress_update) {
  79                 fprintf(stderr, "%4u%% (%u/%u) done\r", percent, n, total);
  80                 progress_update = 0;
  81         }
  82         return percent;
  83 }
  84
  85 /* We always read in 4kB chunks. */
  86 static unsigned char input_buffer[4096];
  87 static unsigned int input_offset, input_len;
  88 static off_t consumed_bytes;
  89 static SHA_CTX input_ctx;
  90 static uint32_t input_crc32;
  91 static int input_fd, output_fd, pack_fd;
  92
  93 /* Discard current buffer used content. */
  94 static void flush(void)
  95 {
  96         if (input_offset) {
  97                 if (output_fd >= 0)
  98                         write_or_die(output_fd, input_buffer, input_offset);
  99                 SHA1_Update(&input_ctx, input_buffer, input_offset);
 100                 memmove(input_buffer, input_buffer + input_offset, input_len);
 101                 input_offset = 0;
 102         }
 103 }
 104
 105 /*
 106  * Make sure at least "min" bytes are available in the buffer, and
 107  * return the pointer to the buffer.
 108  */
 109 static void *fill(int min)
 110 {
 111         if (min <= input_len)
 112                 return input_buffer + input_offset;
 113         if (min > sizeof(input_buffer))
 114                 die("cannot fill %d bytes", min);
 115         flush();
 116         do {
 117                 int ret = xread(input_fd, input_buffer + input_len,
 118                                 sizeof(input_buffer) - input_len);
 119                 if (ret <= 0) {
 120                         if (!ret)
 121                                 die("early EOF");
 122                         die("read error on input: %s", strerror(errno));
 123                 }
 124                 input_len += ret;
 125         } while (input_len < min);
 126         return input_buffer;
 127 }
 128
 129 static void use(int bytes)
 130 {
 131         if (bytes > input_len)
 132                 die("used more bytes than were available");
 133         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 134         input_len -= bytes;
 135         input_offset += bytes;
 136
 137         /* make sure off_t is sufficiently large not to wrap */
 138         if (consumed_bytes > consumed_bytes + bytes)
 139                 die("pack too large for current definition of off_t");
 140         consumed_bytes += bytes;
 141 }
 142
 143 static const char *open_pack_file(const char *pack_name)
 144 {
 145         if (from_stdin) {
 146                 input_fd = 0;
 147                 if (!pack_name) {
 148                         static char tmpfile[PATH_MAX];
 149                         snprintf(tmpfile, sizeof(tmpfile),
 150                                  "%s/tmp_pack_XXXXXX", get_object_directory());
 151                         output_fd = mkstemp(tmpfile);
 152                         pack_name = xstrdup(tmpfile);
 153                 } else
 154                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 155                 if (output_fd < 0)
 156                         die("unable to create %s: %s\n", pack_name, strerror(errno));
 157                 pack_fd = output_fd;
 158         } else {
 159                 input_fd = open(pack_name, O_RDONLY);
 160                 if (input_fd < 0)
 161                         die("cannot open packfile '%s': %s",
 162                             pack_name, strerror(errno));
 163                 output_fd = -1;
 164                 pack_fd = input_fd;
 165         }
 166         SHA1_Init(&input_ctx);
 167         return pack_name;
 168 }
 169
 170 static void parse_pack_header(void)
 171 {
 172         struct pack_header *hdr = fill(sizeof(struct pack_header));
 173
 174         /* Header consistency check */
 175         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 176                 die("pack signature mismatch");
 177         if (!pack_version_ok(hdr->hdr_version))
 178                 die("pack version %d unsupported", ntohl(hdr->hdr_version));
 179
 180         nr_objects = ntohl(hdr->hdr_entries);
 181         use(sizeof(struct pack_header));
 182 }
 183
 184 static void bad_object(unsigned long offset, const char *format,
 185                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 186
 187 static void bad_object(unsigned long offset, const char *format, ...)
 188 {
 189         va_list params;
 190         char buf[1024];
 191
 192         va_start(params, format);
 193         vsnprintf(buf, sizeof(buf), format, params);
 194         va_end(params);
 195         die("pack has bad object at offset %lu: %s", offset, buf);
 196 }
 197
 198 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 199 {
 200         z_stream stream;
 201         void *buf = xmalloc(size);
 202
 203         memset(&stream, 0, sizeof(stream));
 204         stream.next_out = buf;
 205         stream.avail_out = size;
 206         stream.next_in = fill(1);
 207         stream.avail_in = input_len;
 208         inflateInit(&stream);
 209
 210         for (;;) {
 211                 int ret = inflate(&stream, 0);
 212                 use(input_len - stream.avail_in);
 213                 if (stream.total_out == size && ret == Z_STREAM_END)
 214                         break;
 215                 if (ret != Z_OK)
 216                         bad_object(offset, "inflate returned %d", ret);
 217                 stream.next_in = fill(1);
 218                 stream.avail_in = input_len;
 219         }
 220         inflateEnd(&stream);
 221         return buf;
 222 }
 223
 224 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 225 {
 226         unsigned char *p, c;
 227         unsigned long size;
 228         off_t base_offset;
 229         unsigned shift;
 230         void *data;
 231
 232         obj->offset = consumed_bytes;
 233         input_crc32 = crc32(0, Z_NULL, 0);
 234
 235         p = fill(1);
 236         c = *p;
 237         use(1);
 238         obj->type = (c >> 4) & 7;
 239         size = (c & 15);
 240         shift = 4;
 241         while (c & 0x80) {
 242                 p = fill(1);
 243                 c = *p;
 244                 use(1);
 245                 size += (c & 0x7fUL) << shift;
 246                 shift += 7;
 247         }
 248         obj->size = size;
 249
 250         switch (obj->type) {
 251         case OBJ_REF_DELTA:
 252                 hashcpy(delta_base->sha1, fill(20));
 253                 use(20);
 254                 break;
 255         case OBJ_OFS_DELTA:
 256                 memset(delta_base, 0, sizeof(*delta_base));
 257                 p = fill(1);
 258                 c = *p;
 259                 use(1);
 260                 base_offset = c & 127;
 261                 while (c & 128) {
 262                         base_offset += 1;
 263                         if (!base_offset || MSB(base_offset, 7))
 264                                 bad_object(obj->offset, "offset value overflow for delta base object");
 265                         p = fill(1);
 266                         c = *p;
 267                         use(1);
 268                         base_offset = (base_offset << 7) + (c & 127);
 269                 }
 270                 delta_base->offset = obj->offset - base_offset;
 271                 if (delta_base->offset >= obj->offset)
 272                         bad_object(obj->offset, "delta base offset is out of bound");
 273                 break;
 274         case OBJ_COMMIT:
 275         case OBJ_TREE:
 276         case OBJ_BLOB:
 277         case OBJ_TAG:
 278                 break;
 279         default:
 280                 bad_object(obj->offset, "unknown object type %d", obj->type);
 281         }
 282         obj->hdr_size = consumed_bytes - obj->offset;
 283
 284         data = unpack_entry_data(obj->offset, obj->size);
 285         obj->crc32 = input_crc32;
 286         return data;
 287 }
 288
 289 static void *get_data_from_pack(struct object_entry *obj)
 290 {
 291         unsigned long from = obj[0].offset + obj[0].hdr_size;
 292         unsigned long len = obj[1].offset - from;
 293         unsigned long rdy = 0;
 294         unsigned char *src, *data;
 295         z_stream stream;
 296         int st;
 297
 298         src = xmalloc(len);
 299         data = src;
 300         do {
 301                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 302                 if (n <= 0)
 303                         die("cannot pread pack file: %s", strerror(errno));
 304                 rdy += n;
 305         } while (rdy < len);
 306         data = xmalloc(obj->size);
 307         memset(&stream, 0, sizeof(stream));
 308         stream.next_out = data;
 309         stream.avail_out = obj->size;
 310         stream.next_in = src;
 311         stream.avail_in = len;
 312         inflateInit(&stream);
 313         while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
 314         inflateEnd(&stream);
 315         if (st != Z_STREAM_END || stream.total_out != obj->size)
 316                 die("serious inflate inconsistency");
 317         free(src);
 318         return data;
 319 }
 320
 321 static int find_delta(const union delta_base *base)
 322 {
 323         int first = 0, last = nr_deltas;
 324
 325         while (first < last) {
 326                 int next = (first + last) / 2;
 327                 struct delta_entry *delta = &deltas[next];
 328                 int cmp;
 329
 330                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 331                 if (!cmp)
 332                         return next;
 333                 if (cmp < 0) {
 334                         last = next;
 335                         continue;
 336                 }
 337                 first = next+1;
 338         }
 339         return -first-1;
 340 }
 341
 342 static int find_delta_children(const union delta_base *base,
 343                                int *first_index, int *last_index)
 344 {
 345         int first = find_delta(base);
 346         int last = first;
 347         int end = nr_deltas - 1;
 348
 349         if (first < 0)
 350                 return -1;
 351         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 352                 --first;
 353         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 354                 ++last;
 355         *first_index = first;
 356         *last_index = last;
 357         return 0;
 358 }
 359
 360 static void sha1_object(const void *data, unsigned long size,
 361                         enum object_type type, unsigned char *sha1)
 362 {
 363         hash_sha1_file(data, size, typename(type), sha1);
 364         if (has_sha1_file(sha1)) {
 365                 void *has_data;
 366                 enum object_type has_type;
 367                 unsigned long has_size;
 368                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 369                 if (!has_data)
 370                         die("cannot read existing object %s", sha1_to_hex(sha1));
 371                 if (size != has_size || type != has_type ||
 372                     memcmp(data, has_data, size) != 0)
 373                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 374                 free(has_data);
 375         }
 376 }
 377
 378 static void resolve_delta(struct object_entry *delta_obj, void *base_data,
 379                           unsigned long base_size, enum object_type type)
 380 {
 381         void *delta_data;
 382         unsigned long delta_size;
 383         void *result;
 384         unsigned long result_size;
 385         union delta_base delta_base;
 386         int j, first, last;
 387
 388         delta_obj->real_type = type;
 389         delta_data = get_data_from_pack(delta_obj);
 390         delta_size = delta_obj->size;
 391         result = patch_delta(base_data, base_size, delta_data, delta_size,
 392                              &result_size);
 393         free(delta_data);
 394         if (!result)
 395                 bad_object(delta_obj->offset, "failed to apply delta");
 396         sha1_object(result, result_size, type, delta_obj->sha1);
 397         nr_resolved_deltas++;
 398
 399         hashcpy(delta_base.sha1, delta_obj->sha1);
 400         if (!find_delta_children(&delta_base, &first, &last)) {
 401                 for (j = first; j <= last; j++) {
 402                         struct object_entry *child = objects + deltas[j].obj_no;
 403                         if (child->real_type == OBJ_REF_DELTA)
 404                                 resolve_delta(child, result, result_size, type);
 405                 }
 406         }
 407
 408         memset(&delta_base, 0, sizeof(delta_base));
 409         delta_base.offset = delta_obj->offset;
 410         if (!find_delta_children(&delta_base, &first, &last)) {
 411                 for (j = first; j <= last; j++) {
 412                         struct object_entry *child = objects + deltas[j].obj_no;
 413                         if (child->real_type == OBJ_OFS_DELTA)
 414                                 resolve_delta(child, result, result_size, type);
 415                 }
 416         }
 417
 418         free(result);
 419 }
 420
 421 static int compare_delta_entry(const void *a, const void *b)
 422 {
 423         const struct delta_entry *delta_a = a;
 424         const struct delta_entry *delta_b = b;
 425         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 426 }
 427
 428 /* Parse all objects and return the pack content SHA1 hash */
 429 static void parse_pack_objects(unsigned char *sha1)
 430 {
 431         int i, percent = -1;
 432         struct delta_entry *delta = deltas;
 433         void *data;
 434         struct stat st;
 435
 436         /*
 437          * First pass:
 438          * - find locations of all objects;
 439          * - calculate SHA1 of all non-delta objects;
 440          * - remember base (SHA1 or offset) for all deltas.
 441          */
 442         if (verbose)
 443                 fprintf(stderr, "Indexing %d objects.\n", nr_objects);
 444         for (i = 0; i < nr_objects; i++) {
 445                 struct object_entry *obj = &objects[i];
 446                 data = unpack_raw_entry(obj, &delta->base);
 447                 obj->real_type = obj->type;
 448                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 449                         nr_deltas++;
 450                         delta->obj_no = i;
 451                         delta++;
 452                 } else
 453                         sha1_object(data, obj->size, obj->type, obj->sha1);
 454                 free(data);
 455                 if (verbose)
 456                         percent = display_progress(i+1, nr_objects, percent);
 457         }
 458         objects[i].offset = consumed_bytes;
 459         if (verbose)
 460                 fputc('\n', stderr);
 461
 462         /* Check pack integrity */
 463         flush();
 464         SHA1_Final(sha1, &input_ctx);
 465         if (hashcmp(fill(20), sha1))
 466                 die("pack is corrupted (SHA1 mismatch)");
 467         use(20);
 468
 469         /* If input_fd is a file, we should have reached its end now. */
 470         if (fstat(input_fd, &st))
 471                 die("cannot fstat packfile: %s", strerror(errno));
 472         if (S_ISREG(st.st_mode) &&
 473                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 474                 die("pack has junk at the end");
 475
 476         if (!nr_deltas)
 477                 return;
 478
 479         /* Sort deltas by base SHA1/offset for fast searching */
 480         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 481               compare_delta_entry);
 482
 483         /*
 484          * Second pass:
 485          * - for all non-delta objects, look if it is used as a base for
 486          *   deltas;
 487          * - if used as a base, uncompress the object and apply all deltas,
 488          *   recursively checking if the resulting object is used as a base
 489          *   for some more deltas.
 490          */
 491         if (verbose)
 492                 fprintf(stderr, "Resolving %d deltas.\n", nr_deltas);
 493         for (i = 0; i < nr_objects; i++) {
 494                 struct object_entry *obj = &objects[i];
 495                 union delta_base base;
 496                 int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 497
 498                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 499                         continue;
 500                 hashcpy(base.sha1, obj->sha1);
 501                 ref = !find_delta_children(&base, &ref_first, &ref_last);
 502                 memset(&base, 0, sizeof(base));
 503                 base.offset = obj->offset;
 504                 ofs = !find_delta_children(&base, &ofs_first, &ofs_last);
 505                 if (!ref && !ofs)
 506                         continue;
 507                 data = get_data_from_pack(obj);
 508                 if (ref)
 509                         for (j = ref_first; j <= ref_last; j++) {
 510                                 struct object_entry *child = objects + deltas[j].obj_no;
 511                                 if (child->real_type == OBJ_REF_DELTA)
 512                                         resolve_delta(child, data,
 513                                                       obj->size, obj->type);
 514                         }
 515                 if (ofs)
 516                         for (j = ofs_first; j <= ofs_last; j++) {
 517                                 struct object_entry *child = objects + deltas[j].obj_no;
 518                                 if (child->real_type == OBJ_OFS_DELTA)
 519                                         resolve_delta(child, data,
 520                                                       obj->size, obj->type);
 521                         }
 522                 free(data);
 523                 if (verbose)
 524                         percent = display_progress(nr_resolved_deltas,
 525                                                    nr_deltas, percent);
 526         }
 527         if (verbose && nr_resolved_deltas == nr_deltas)
 528                 fputc('\n', stderr);
 529 }
 530
 531 static int write_compressed(int fd, void *in, unsigned int size, uint32_t *obj_crc)
 532 {
 533         z_stream stream;
 534         unsigned long maxsize;
 535         void *out;
 536
 537         memset(&stream, 0, sizeof(stream));
 538         deflateInit(&stream, zlib_compression_level);
 539         maxsize = deflateBound(&stream, size);
 540         out = xmalloc(maxsize);
 541
 542         /* Compress it */
 543         stream.next_in = in;
 544         stream.avail_in = size;
 545         stream.next_out = out;
 546         stream.avail_out = maxsize;
 547         while (deflate(&stream, Z_FINISH) == Z_OK);
 548         deflateEnd(&stream);
 549
 550         size = stream.total_out;
 551         write_or_die(fd, out, size);
 552         *obj_crc = crc32(*obj_crc, out, size);
 553         free(out);
 554         return size;
 555 }
 556
 557 static void append_obj_to_pack(const unsigned char *sha1, void *buf,
 558                                unsigned long size, enum object_type type)
 559 {
 560         struct object_entry *obj = &objects[nr_objects++];
 561         unsigned char header[10];
 562         unsigned long s = size;
 563         int n = 0;
 564         unsigned char c = (type << 4) | (s & 15);
 565         s >>= 4;
 566         while (s) {
 567                 header[n++] = c | 0x80;
 568                 c = s & 0x7f;
 569                 s >>= 7;
 570         }
 571         header[n++] = c;
 572         write_or_die(output_fd, header, n);
 573         obj[0].crc32 = crc32(0, Z_NULL, 0);
 574         obj[0].crc32 = crc32(obj[0].crc32, header, n);
 575         obj[1].offset = obj[0].offset + n;
 576         obj[1].offset += write_compressed(output_fd, buf, size, &obj[0].crc32);
 577         hashcpy(obj->sha1, sha1);
 578 }
 579
 580 static int delta_pos_compare(const void *_a, const void *_b)
 581 {
 582         struct delta_entry *a = *(struct delta_entry **)_a;
 583         struct delta_entry *b = *(struct delta_entry **)_b;
 584         return a->obj_no - b->obj_no;
 585 }
 586
 587 static void fix_unresolved_deltas(int nr_unresolved)
 588 {
 589         struct delta_entry **sorted_by_pos;
 590         int i, n = 0, percent = -1;
 591
 592         /*
 593          * Since many unresolved deltas may well be themselves base objects
 594          * for more unresolved deltas, we really want to include the
 595          * smallest number of base objects that would cover as much delta
 596          * as possible by picking the
 597          * trunc deltas first, allowing for other deltas to resolve without
 598          * additional base objects.  Since most base objects are to be found
 599          * before deltas depending on them, a good heuristic is to start
 600          * resolving deltas in the same order as their position in the pack.
 601          */
 602         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 603         for (i = 0; i < nr_deltas; i++) {
 604                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 605                         continue;
 606                 sorted_by_pos[n++] = &deltas[i];
 607         }
 608         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 609
 610         for (i = 0; i < n; i++) {
 611                 struct delta_entry *d = sorted_by_pos[i];
 612                 void *data;
 613                 unsigned long size;
 614                 enum object_type type;
 615                 int j, first, last;
 616
 617                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 618                         continue;
 619                 data = read_sha1_file(d->base.sha1, &type, &size);
 620                 if (!data)
 621                         continue;
 622
 623                 find_delta_children(&d->base, &first, &last);
 624                 for (j = first; j <= last; j++) {
 625                         struct object_entry *child = objects + deltas[j].obj_no;
 626                         if (child->real_type == OBJ_REF_DELTA)
 627                                 resolve_delta(child, data, size, type);
 628                 }
 629
 630                 if (check_sha1_signature(d->base.sha1, data, size, typename(type)))
 631                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 632                 append_obj_to_pack(d->base.sha1, data, size, type);
 633                 free(data);
 634                 if (verbose)
 635                         percent = display_progress(nr_resolved_deltas,
 636                                                    nr_deltas, percent);
 637         }
 638         free(sorted_by_pos);
 639         if (verbose)
 640                 fputc('\n', stderr);
 641 }
 642
 643 static void readjust_pack_header_and_sha1(unsigned char *sha1)
 644 {
 645         struct pack_header hdr;
 646         SHA_CTX ctx;
 647         int size;
 648
 649         /* Rewrite pack header with updated object number */
 650         if (lseek(output_fd, 0, SEEK_SET) != 0)
 651                 die("cannot seek back: %s", strerror(errno));
 652         if (read_in_full(output_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 653                 die("cannot read pack header back: %s", strerror(errno));
 654         hdr.hdr_entries = htonl(nr_objects);
 655         if (lseek(output_fd, 0, SEEK_SET) != 0)
 656                 die("cannot seek back: %s", strerror(errno));
 657         write_or_die(output_fd, &hdr, sizeof(hdr));
 658         if (lseek(output_fd, 0, SEEK_SET) != 0)
 659                 die("cannot seek back: %s", strerror(errno));
 660
 661         /* Recompute and store the new pack's SHA1 */
 662         SHA1_Init(&ctx);
 663         do {
 664                 unsigned char *buf[4096];
 665                 size = xread(output_fd, buf, sizeof(buf));
 666                 if (size < 0)
 667                         die("cannot read pack data back: %s", strerror(errno));
 668                 SHA1_Update(&ctx, buf, size);
 669         } while (size > 0);
 670         SHA1_Final(sha1, &ctx);
 671         write_or_die(output_fd, sha1, 20);
 672 }
 673
 674 static uint32_t index_default_version = 1;
 675 static uint32_t index_off32_limit = 0x7fffffff;
 676
 677 static int sha1_compare(const void *_a, const void *_b)
 678 {
 679         struct object_entry *a = *(struct object_entry **)_a;
 680         struct object_entry *b = *(struct object_entry **)_b;
 681         return hashcmp(a->sha1, b->sha1);
 682 }
 683
 684 /*
 685  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 686  * the SHA1 hash of sorted object names.
 687  */
 688 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 689 {
 690         struct sha1file *f;
 691         struct object_entry **sorted_by_sha, **list, **last;
 692         uint32_t array[256];
 693         int i, fd;
 694         SHA_CTX ctx;
 695         uint32_t index_version;
 696
 697         if (nr_objects) {
 698                 sorted_by_sha =
 699                         xcalloc(nr_objects, sizeof(struct object_entry *));
 700                 list = sorted_by_sha;
 701                 last = sorted_by_sha + nr_objects;
 702                 for (i = 0; i < nr_objects; ++i)
 703                         sorted_by_sha[i] = &objects[i];
 704                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 705                       sha1_compare);
 706         }
 707         else
 708                 sorted_by_sha = list = last = NULL;
 709
 710         if (!index_name) {
 711                 static char tmpfile[PATH_MAX];
 712                 snprintf(tmpfile, sizeof(tmpfile),
 713                          "%s/tmp_idx_XXXXXX", get_object_directory());
 714                 fd = mkstemp(tmpfile);
 715                 index_name = xstrdup(tmpfile);
 716         } else {
 717                 unlink(index_name);
 718                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 719         }
 720         if (fd < 0)
 721                 die("unable to create %s: %s", index_name, strerror(errno));
 722         f = sha1fd(fd, index_name);
 723
 724         /* if last object's offset is >= 2^31 we should use index V2 */
 725         index_version = (objects[nr_objects-1].offset >> 31) ? 2 : index_default_version;
 726
 727         /* index versions 2 and above need a header */
 728         if (index_version >= 2) {
 729                 struct pack_idx_header hdr;
 730                 hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
 731                 hdr.idx_version = htonl(index_version);
 732                 sha1write(f, &hdr, sizeof(hdr));
 733         }
 734
 735         /*
 736          * Write the first-level table (the list is sorted,
 737          * but we use a 256-entry lookup to be able to avoid
 738          * having to do eight extra binary search iterations).
 739          */
 740         for (i = 0; i < 256; i++) {
 741                 struct object_entry **next = list;
 742                 while (next < last) {
 743                         struct object_entry *obj = *next;
 744                         if (obj->sha1[0] != i)
 745                                 break;
 746                         next++;
 747                 }
 748                 array[i] = htonl(next - sorted_by_sha);
 749                 list = next;
 750         }
 751         sha1write(f, array, 256 * 4);
 752
 753         /* compute the SHA1 hash of sorted object names. */
 754         SHA1_Init(&ctx);
 755
 756         /*
 757          * Write the actual SHA1 entries..
 758          */
 759         list = sorted_by_sha;
 760         for (i = 0; i < nr_objects; i++) {
 761                 struct object_entry *obj = *list++;
 762                 if (index_version < 2) {
 763                         uint32_t offset = htonl(obj->offset);
 764                         sha1write(f, &offset, 4);
 765                 }
 766                 sha1write(f, obj->sha1, 20);
 767                 SHA1_Update(&ctx, obj->sha1, 20);
 768         }
 769
 770         if (index_version >= 2) {
 771                 unsigned int nr_large_offset = 0;
 772
 773                 /* write the crc32 table */
 774                 list = sorted_by_sha;
 775                 for (i = 0; i < nr_objects; i++) {
 776                         struct object_entry *obj = *list++;
 777                         uint32_t crc32_val = htonl(obj->crc32);
 778                         sha1write(f, &crc32_val, 4);
 779                 }
 780
 781                 /* write the 32-bit offset table */
 782                 list = sorted_by_sha;
 783                 for (i = 0; i < nr_objects; i++) {
 784                         struct object_entry *obj = *list++;
 785                         uint32_t offset = (obj->offset <= index_off32_limit) ?
 786                                 obj->offset : (0x80000000 | nr_large_offset++);
 787                         offset = htonl(offset);
 788                         sha1write(f, &offset, 4);
 789                 }
 790
 791                 /* write the large offset table */
 792                 list = sorted_by_sha;
 793                 while (nr_large_offset) {
 794                         struct object_entry *obj = *list++;
 795                         uint64_t offset = obj->offset;
 796                         if (offset > index_off32_limit) {
 797                                 uint32_t split[2];
 798                                 split[0]        = htonl(offset >> 32);
 799                                 split[1] = htonl(offset & 0xffffffff);
 800                                 sha1write(f, split, 8);
 801                                 nr_large_offset--;
 802                         }
 803                 }
 804         }
 805
 806         sha1write(f, sha1, 20);
 807         sha1close(f, NULL, 1);
 808         free(sorted_by_sha);
 809         SHA1_Final(sha1, &ctx);
 810         return index_name;
 811 }
 812
 813 static void final(const char *final_pack_name, const char *curr_pack_name,
 814                   const char *final_index_name, const char *curr_index_name,
 815                   const char *keep_name, const char *keep_msg,
 816                   unsigned char *sha1)
 817 {
 818         const char *report = "pack";
 819         char name[PATH_MAX];
 820         int err;
 821
 822         if (!from_stdin) {
 823                 close(input_fd);
 824         } else {
 825                 err = close(output_fd);
 826                 if (err)
 827                         die("error while closing pack file: %s", strerror(errno));
 828                 chmod(curr_pack_name, 0444);
 829         }
 830
 831         if (keep_msg) {
 832                 int keep_fd, keep_msg_len = strlen(keep_msg);
 833                 if (!keep_name) {
 834                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 835                                  get_object_directory(), sha1_to_hex(sha1));
 836                         keep_name = name;
 837                 }
 838                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 839                 if (keep_fd < 0) {
 840                         if (errno != EEXIST)
 841                                 die("cannot write keep file");
 842                 } else {
 843                         if (keep_msg_len > 0) {
 844                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 845                                 write_or_die(keep_fd, "\n", 1);
 846                         }
 847                         close(keep_fd);
 848                         report = "keep";
 849                 }
 850         }
 851
 852         if (final_pack_name != curr_pack_name) {
 853                 if (!final_pack_name) {
 854                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 855                                  get_object_directory(), sha1_to_hex(sha1));
 856                         final_pack_name = name;
 857                 }
 858                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 859                         die("cannot store pack file");
 860         }
 861
 862         chmod(curr_index_name, 0444);
 863         if (final_index_name != curr_index_name) {
 864                 if (!final_index_name) {
 865                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 866                                  get_object_directory(), sha1_to_hex(sha1));
 867                         final_index_name = name;
 868                 }
 869                 if (move_temp_to_file(curr_index_name, final_index_name))
 870                         die("cannot store index file");
 871         }
 872
 873         if (!from_stdin) {
 874                 printf("%s\n", sha1_to_hex(sha1));
 875         } else {
 876                 char buf[48];
 877                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 878                                    report, sha1_to_hex(sha1));
 879                 write_or_die(1, buf, len);
 880
 881                 /*
 882                  * Let's just mimic git-unpack-objects here and write
 883                  * the last part of the input buffer to stdout.
 884                  */
 885                 while (input_len) {
 886                         err = xwrite(1, input_buffer + input_offset, input_len);
 887                         if (err <= 0)
 888                                 break;
 889                         input_len -= err;
 890                         input_offset += err;
 891                 }
 892         }
 893 }
 894
 895 int main(int argc, char **argv)
 896 {
 897         int i, fix_thin_pack = 0;
 898         const char *curr_pack, *pack_name = NULL;
 899         const char *curr_index, *index_name = NULL;
 900         const char *keep_name = NULL, *keep_msg = NULL;
 901         char *index_name_buf = NULL, *keep_name_buf = NULL;
 902         unsigned char sha1[20];
 903
 904         for (i = 1; i < argc; i++) {
 905                 const char *arg = argv[i];
 906
 907                 if (*arg == '-') {
 908                         if (!strcmp(arg, "--stdin")) {
 909                                 from_stdin = 1;
 910                         } else if (!strcmp(arg, "--fix-thin")) {
 911                                 fix_thin_pack = 1;
 912                         } else if (!strcmp(arg, "--keep")) {
 913                                 keep_msg = "";
 914                         } else if (!prefixcmp(arg, "--keep=")) {
 915                                 keep_msg = arg + 7;
 916                         } else if (!prefixcmp(arg, "--pack_header=")) {
 917                                 struct pack_header *hdr;
 918                                 char *c;
 919
 920                                 hdr = (struct pack_header *)input_buffer;
 921                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 922                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 923                                 if (*c != ',')
 924                                         die("bad %s", arg);
 925                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 926                                 if (*c)
 927                                         die("bad %s", arg);
 928                                 input_len = sizeof(*hdr);
 929                         } else if (!strcmp(arg, "-v")) {
 930                                 verbose = 1;
 931                         } else if (!strcmp(arg, "-o")) {
 932                                 if (index_name || (i+1) >= argc)
 933                                         usage(index_pack_usage);
 934                                 index_name = argv[++i];
 935                         } else if (!prefixcmp(arg, "--index-version=")) {
 936                                 char *c;
 937                                 index_default_version = strtoul(arg + 16, &c, 10);
 938                                 if (index_default_version > 2)
 939                                         die("bad %s", arg);
 940                                 if (*c == ',')
 941                                         index_off32_limit = strtoul(c+1, &c, 0);
 942                                 if (*c || index_off32_limit & 0x80000000)
 943                                         die("bad %s", arg);
 944                         } else
 945                                 usage(index_pack_usage);
 946                         continue;
 947                 }
 948
 949                 if (pack_name)
 950                         usage(index_pack_usage);
 951                 pack_name = arg;
 952         }
 953
 954         if (!pack_name && !from_stdin)
 955                 usage(index_pack_usage);
 956         if (fix_thin_pack && !from_stdin)
 957                 die("--fix-thin cannot be used without --stdin");
 958         if (!index_name && pack_name) {
 959                 int len = strlen(pack_name);
 960                 if (!has_extension(pack_name, ".pack"))
 961                         die("packfile name '%s' does not end with '.pack'",
 962                             pack_name);
 963                 index_name_buf = xmalloc(len);
 964                 memcpy(index_name_buf, pack_name, len - 5);
 965                 strcpy(index_name_buf + len - 5, ".idx");
 966                 index_name = index_name_buf;
 967         }
 968         if (keep_msg && !keep_name && pack_name) {
 969                 int len = strlen(pack_name);
 970                 if (!has_extension(pack_name, ".pack"))
 971                         die("packfile name '%s' does not end with '.pack'",
 972                             pack_name);
 973                 keep_name_buf = xmalloc(len);
 974                 memcpy(keep_name_buf, pack_name, len - 5);
 975                 strcpy(keep_name_buf + len - 5, ".keep");
 976                 keep_name = keep_name_buf;
 977         }
 978
 979         curr_pack = open_pack_file(pack_name);
 980         parse_pack_header();
 981         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 982         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 983         if (verbose)
 984                 setup_progress_signal();
 985         parse_pack_objects(sha1);
 986         if (nr_deltas != nr_resolved_deltas) {
 987                 if (fix_thin_pack) {
 988                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 989                         int nr_objects_initial = nr_objects;
 990                         if (nr_unresolved <= 0)
 991                                 die("confusion beyond insanity");
 992                         objects = xrealloc(objects,
 993                                            (nr_objects + nr_unresolved + 1)
 994                                            * sizeof(*objects));
 995                         fix_unresolved_deltas(nr_unresolved);
 996                         if (verbose)
 997                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 998                                         nr_objects - nr_objects_initial);
 999                         readjust_pack_header_and_sha1(sha1);
1000                 }
1001                 if (nr_deltas != nr_resolved_deltas)
1002                         die("pack has %d unresolved deltas",
1003                             nr_deltas - nr_resolved_deltas);
1004         } else {
1005                 /* Flush remaining pack final 20-byte SHA1. */
1006                 flush();
1007         }
1008         free(deltas);
1009         curr_index = write_index_file(index_name, sha1);
1010         final(pack_name, curr_pack,
1011                 index_name, curr_index,
1012                 keep_name, keep_msg,
1013                 sha1);
1014         free(objects);
1015         free(index_name_buf);
1016         free(keep_name_buf);
1017
1018         return 0;
1019 }