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 int sha1_compare(const void *_a, const void *_b)
 675 {
 676         struct object_entry *a = *(struct object_entry **)_a;
 677         struct object_entry *b = *(struct object_entry **)_b;
 678         return hashcmp(a->sha1, b->sha1);
 679 }
 680
 681 /*
 682  * On entry *sha1 contains the pack content SHA1 hash, on exit it is
 683  * the SHA1 hash of sorted object names.
 684  */
 685 static const char *write_index_file(const char *index_name, unsigned char *sha1)
 686 {
 687         struct sha1file *f;
 688         struct object_entry **sorted_by_sha, **list, **last;
 689         unsigned int array[256];
 690         int i, fd;
 691         SHA_CTX ctx;
 692
 693         if (nr_objects) {
 694                 sorted_by_sha =
 695                         xcalloc(nr_objects, sizeof(struct object_entry *));
 696                 list = sorted_by_sha;
 697                 last = sorted_by_sha + nr_objects;
 698                 for (i = 0; i < nr_objects; ++i)
 699                         sorted_by_sha[i] = &objects[i];
 700                 qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
 701                       sha1_compare);
 702
 703         }
 704         else
 705                 sorted_by_sha = list = last = NULL;
 706
 707         if (!index_name) {
 708                 static char tmpfile[PATH_MAX];
 709                 snprintf(tmpfile, sizeof(tmpfile),
 710                          "%s/tmp_idx_XXXXXX", get_object_directory());
 711                 fd = mkstemp(tmpfile);
 712                 index_name = xstrdup(tmpfile);
 713         } else {
 714                 unlink(index_name);
 715                 fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 716         }
 717         if (fd < 0)
 718                 die("unable to create %s: %s", index_name, strerror(errno));
 719         f = sha1fd(fd, index_name);
 720
 721         /*
 722          * Write the first-level table (the list is sorted,
 723          * but we use a 256-entry lookup to be able to avoid
 724          * having to do eight extra binary search iterations).
 725          */
 726         for (i = 0; i < 256; i++) {
 727                 struct object_entry **next = list;
 728                 while (next < last) {
 729                         struct object_entry *obj = *next;
 730                         if (obj->sha1[0] != i)
 731                                 break;
 732                         next++;
 733                 }
 734                 array[i] = htonl(next - sorted_by_sha);
 735                 list = next;
 736         }
 737         sha1write(f, array, 256 * sizeof(int));
 738
 739         /* recompute the SHA1 hash of sorted object names.
 740          * currently pack-objects does not do this, but that
 741          * can be fixed.
 742          */
 743         SHA1_Init(&ctx);
 744         /*
 745          * Write the actual SHA1 entries..
 746          */
 747         list = sorted_by_sha;
 748         for (i = 0; i < nr_objects; i++) {
 749                 struct object_entry *obj = *list++;
 750                 unsigned int offset = htonl(obj->offset);
 751                 sha1write(f, &offset, 4);
 752                 sha1write(f, obj->sha1, 20);
 753                 SHA1_Update(&ctx, obj->sha1, 20);
 754         }
 755         sha1write(f, sha1, 20);
 756         sha1close(f, NULL, 1);
 757         free(sorted_by_sha);
 758         SHA1_Final(sha1, &ctx);
 759         return index_name;
 760 }
 761
 762 static void final(const char *final_pack_name, const char *curr_pack_name,
 763                   const char *final_index_name, const char *curr_index_name,
 764                   const char *keep_name, const char *keep_msg,
 765                   unsigned char *sha1)
 766 {
 767         const char *report = "pack";
 768         char name[PATH_MAX];
 769         int err;
 770
 771         if (!from_stdin) {
 772                 close(input_fd);
 773         } else {
 774                 err = close(output_fd);
 775                 if (err)
 776                         die("error while closing pack file: %s", strerror(errno));
 777                 chmod(curr_pack_name, 0444);
 778         }
 779
 780         if (keep_msg) {
 781                 int keep_fd, keep_msg_len = strlen(keep_msg);
 782                 if (!keep_name) {
 783                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 784                                  get_object_directory(), sha1_to_hex(sha1));
 785                         keep_name = name;
 786                 }
 787                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 788                 if (keep_fd < 0) {
 789                         if (errno != EEXIST)
 790                                 die("cannot write keep file");
 791                 } else {
 792                         if (keep_msg_len > 0) {
 793                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 794                                 write_or_die(keep_fd, "\n", 1);
 795                         }
 796                         close(keep_fd);
 797                         report = "keep";
 798                 }
 799         }
 800
 801         if (final_pack_name != curr_pack_name) {
 802                 if (!final_pack_name) {
 803                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 804                                  get_object_directory(), sha1_to_hex(sha1));
 805                         final_pack_name = name;
 806                 }
 807                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 808                         die("cannot store pack file");
 809         }
 810
 811         chmod(curr_index_name, 0444);
 812         if (final_index_name != curr_index_name) {
 813                 if (!final_index_name) {
 814                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 815                                  get_object_directory(), sha1_to_hex(sha1));
 816                         final_index_name = name;
 817                 }
 818                 if (move_temp_to_file(curr_index_name, final_index_name))
 819                         die("cannot store index file");
 820         }
 821
 822         if (!from_stdin) {
 823                 printf("%s\n", sha1_to_hex(sha1));
 824         } else {
 825                 char buf[48];
 826                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 827                                    report, sha1_to_hex(sha1));
 828                 write_or_die(1, buf, len);
 829
 830                 /*
 831                  * Let's just mimic git-unpack-objects here and write
 832                  * the last part of the input buffer to stdout.
 833                  */
 834                 while (input_len) {
 835                         err = xwrite(1, input_buffer + input_offset, input_len);
 836                         if (err <= 0)
 837                                 break;
 838                         input_len -= err;
 839                         input_offset += err;
 840                 }
 841         }
 842 }
 843
 844 int main(int argc, char **argv)
 845 {
 846         int i, fix_thin_pack = 0;
 847         const char *curr_pack, *pack_name = NULL;
 848         const char *curr_index, *index_name = NULL;
 849         const char *keep_name = NULL, *keep_msg = NULL;
 850         char *index_name_buf = NULL, *keep_name_buf = NULL;
 851         unsigned char sha1[20];
 852
 853         for (i = 1; i < argc; i++) {
 854                 const char *arg = argv[i];
 855
 856                 if (*arg == '-') {
 857                         if (!strcmp(arg, "--stdin")) {
 858                                 from_stdin = 1;
 859                         } else if (!strcmp(arg, "--fix-thin")) {
 860                                 fix_thin_pack = 1;
 861                         } else if (!strcmp(arg, "--keep")) {
 862                                 keep_msg = "";
 863                         } else if (!prefixcmp(arg, "--keep=")) {
 864                                 keep_msg = arg + 7;
 865                         } else if (!prefixcmp(arg, "--pack_header=")) {
 866                                 struct pack_header *hdr;
 867                                 char *c;
 868
 869                                 hdr = (struct pack_header *)input_buffer;
 870                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 871                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 872                                 if (*c != ',')
 873                                         die("bad %s", arg);
 874                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 875                                 if (*c)
 876                                         die("bad %s", arg);
 877                                 input_len = sizeof(*hdr);
 878                         } else if (!strcmp(arg, "-v")) {
 879                                 verbose = 1;
 880                         } else if (!strcmp(arg, "-o")) {
 881                                 if (index_name || (i+1) >= argc)
 882                                         usage(index_pack_usage);
 883                                 index_name = argv[++i];
 884                         } else
 885                                 usage(index_pack_usage);
 886                         continue;
 887                 }
 888
 889                 if (pack_name)
 890                         usage(index_pack_usage);
 891                 pack_name = arg;
 892         }
 893
 894         if (!pack_name && !from_stdin)
 895                 usage(index_pack_usage);
 896         if (fix_thin_pack && !from_stdin)
 897                 die("--fix-thin cannot be used without --stdin");
 898         if (!index_name && pack_name) {
 899                 int len = strlen(pack_name);
 900                 if (!has_extension(pack_name, ".pack"))
 901                         die("packfile name '%s' does not end with '.pack'",
 902                             pack_name);
 903                 index_name_buf = xmalloc(len);
 904                 memcpy(index_name_buf, pack_name, len - 5);
 905                 strcpy(index_name_buf + len - 5, ".idx");
 906                 index_name = index_name_buf;
 907         }
 908         if (keep_msg && !keep_name && pack_name) {
 909                 int len = strlen(pack_name);
 910                 if (!has_extension(pack_name, ".pack"))
 911                         die("packfile name '%s' does not end with '.pack'",
 912                             pack_name);
 913                 keep_name_buf = xmalloc(len);
 914                 memcpy(keep_name_buf, pack_name, len - 5);
 915                 strcpy(keep_name_buf + len - 5, ".keep");
 916                 keep_name = keep_name_buf;
 917         }
 918
 919         curr_pack = open_pack_file(pack_name);
 920         parse_pack_header();
 921         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 922         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 923         if (verbose)
 924                 setup_progress_signal();
 925         parse_pack_objects(sha1);
 926         if (nr_deltas != nr_resolved_deltas) {
 927                 if (fix_thin_pack) {
 928                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 929                         int nr_objects_initial = nr_objects;
 930                         if (nr_unresolved <= 0)
 931                                 die("confusion beyond insanity");
 932                         objects = xrealloc(objects,
 933                                            (nr_objects + nr_unresolved + 1)
 934                                            * sizeof(*objects));
 935                         fix_unresolved_deltas(nr_unresolved);
 936                         if (verbose)
 937                                 fprintf(stderr, "%d objects were added to complete this thin pack.\n",
 938                                         nr_objects - nr_objects_initial);
 939                         readjust_pack_header_and_sha1(sha1);
 940                 }
 941                 if (nr_deltas != nr_resolved_deltas)
 942                         die("pack has %d unresolved deltas",
 943                             nr_deltas - nr_resolved_deltas);
 944         } else {
 945                 /* Flush remaining pack final 20-byte SHA1. */
 946                 flush();
 947         }
 948         free(deltas);
 949         curr_index = write_index_file(index_name, sha1);
 950         final(pack_name, curr_pack,
 951                 index_name, curr_index,
 952                 keep_name, keep_msg,
 953                 sha1);
 954         free(objects);
 955         free(index_name_buf);
 956         free(keep_name_buf);
 957
 958         return 0;
 959 }