builtin/index-pack.c

   1 #include "builtin.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12
  13 static const char index_pack_usage[] =
  14 "git index-pack [-v] [-o <index-file>] [--keep | --keep=<msg>] [--verify] [--strict] (<pack-file> | --stdin [--fix-thin] [<pack-file>])";
  15
  16 struct object_entry {
  17         struct pack_idx_entry idx;
  18         unsigned long size;
  19         unsigned int hdr_size;
  20         enum object_type type;
  21         enum object_type real_type;
  22         unsigned delta_depth;
  23         int base_object_no;
  24 };
  25
  26 union delta_base {
  27         unsigned char sha1[20];
  28         off_t offset;
  29 };
  30
  31 struct base_data {
  32         struct base_data *base;
  33         struct base_data *child;
  34         struct object_entry *obj;
  35         void *data;
  36         unsigned long size;
  37         int ref_first, ref_last;
  38         int ofs_first, ofs_last;
  39 };
  40
  41 /*
  42  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  43  * to memcmp() only the first 20 bytes.
  44  */
  45 #define UNION_BASE_SZ   20
  46
  47 #define FLAG_LINK (1u<<20)
  48 #define FLAG_CHECKED (1u<<21)
  49
  50 struct delta_entry {
  51         union delta_base base;
  52         int obj_no;
  53 };
  54
  55 static struct object_entry *objects;
  56 static struct delta_entry *deltas;
  57 static struct base_data *base_cache;
  58 static size_t base_cache_used;
  59 static int nr_objects;
  60 static int nr_deltas;
  61 static int nr_resolved_deltas;
  62
  63 static int from_stdin;
  64 static int strict;
  65 static int verbose;
  66
  67 static struct progress *progress;
  68
  69 /* We always read in 4kB chunks. */
  70 static unsigned char input_buffer[4096];
  71 static unsigned int input_offset, input_len;
  72 static off_t consumed_bytes;
  73 static unsigned deepest_delta;
  74 static git_SHA_CTX input_ctx;
  75 static uint32_t input_crc32;
  76 static int input_fd, output_fd, pack_fd;
  77
  78 static int mark_link(struct object *obj, int type, void *data)
  79 {
  80         if (!obj)
  81                 return -1;
  82
  83         if (type != OBJ_ANY && obj->type != type)
  84                 die(_("object type mismatch at %s"), sha1_to_hex(obj->sha1));
  85
  86         obj->flags |= FLAG_LINK;
  87         return 0;
  88 }
  89
  90 /* The content of each linked object must have been checked
  91    or it must be already present in the object database */
  92 static void check_object(struct object *obj)
  93 {
  94         if (!obj)
  95                 return;
  96
  97         if (!(obj->flags & FLAG_LINK))
  98                 return;
  99
 100         if (!(obj->flags & FLAG_CHECKED)) {
 101                 unsigned long size;
 102                 int type = sha1_object_info(obj->sha1, &size);
 103                 if (type != obj->type || type <= 0)
 104                         die(_("object of unexpected type"));
 105                 obj->flags |= FLAG_CHECKED;
 106                 return;
 107         }
 108 }
 109
 110 static void check_objects(void)
 111 {
 112         unsigned i, max;
 113
 114         max = get_max_object_index();
 115         for (i = 0; i < max; i++)
 116                 check_object(get_indexed_object(i));
 117 }
 118
 119
 120 /* Discard current buffer used content. */
 121 static void flush(void)
 122 {
 123         if (input_offset) {
 124                 if (output_fd >= 0)
 125                         write_or_die(output_fd, input_buffer, input_offset);
 126                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 127                 memmove(input_buffer, input_buffer + input_offset, input_len);
 128                 input_offset = 0;
 129         }
 130 }
 131
 132 /*
 133  * Make sure at least "min" bytes are available in the buffer, and
 134  * return the pointer to the buffer.
 135  */
 136 static void *fill(int min)
 137 {
 138         if (min <= input_len)
 139                 return input_buffer + input_offset;
 140         if (min > sizeof(input_buffer))
 141                 die(Q_("cannot fill %d byte",
 142                        "cannot fill %d bytes",
 143                        min),
 144                     min);
 145         flush();
 146         do {
 147                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 148                                 sizeof(input_buffer) - input_len);
 149                 if (ret <= 0) {
 150                         if (!ret)
 151                                 die(_("early EOF"));
 152                         die_errno(_("read error on input"));
 153                 }
 154                 input_len += ret;
 155                 if (from_stdin)
 156                         display_throughput(progress, consumed_bytes + input_len);
 157         } while (input_len < min);
 158         return input_buffer;
 159 }
 160
 161 static void use(int bytes)
 162 {
 163         if (bytes > input_len)
 164                 die(_("used more bytes than were available"));
 165         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 166         input_len -= bytes;
 167         input_offset += bytes;
 168
 169         /* make sure off_t is sufficiently large not to wrap */
 170         if (signed_add_overflows(consumed_bytes, bytes))
 171                 die(_("pack too large for current definition of off_t"));
 172         consumed_bytes += bytes;
 173 }
 174
 175 static const char *open_pack_file(const char *pack_name)
 176 {
 177         if (from_stdin) {
 178                 input_fd = 0;
 179                 if (!pack_name) {
 180                         static char tmp_file[PATH_MAX];
 181                         output_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 182                                                 "pack/tmp_pack_XXXXXX");
 183                         pack_name = xstrdup(tmp_file);
 184                 } else
 185                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 186                 if (output_fd < 0)
 187                         die_errno(_("unable to create '%s'"), pack_name);
 188                 pack_fd = output_fd;
 189         } else {
 190                 input_fd = open(pack_name, O_RDONLY);
 191                 if (input_fd < 0)
 192                         die_errno(_("cannot open packfile '%s'"), pack_name);
 193                 output_fd = -1;
 194                 pack_fd = input_fd;
 195         }
 196         git_SHA1_Init(&input_ctx);
 197         return pack_name;
 198 }
 199
 200 static void parse_pack_header(void)
 201 {
 202         struct pack_header *hdr = fill(sizeof(struct pack_header));
 203
 204         /* Header consistency check */
 205         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 206                 die(_("pack signature mismatch"));
 207         if (!pack_version_ok(hdr->hdr_version))
 208                 die("pack version %"PRIu32" unsupported",
 209                         ntohl(hdr->hdr_version));
 210
 211         nr_objects = ntohl(hdr->hdr_entries);
 212         use(sizeof(struct pack_header));
 213 }
 214
 215 static NORETURN void bad_object(unsigned long offset, const char *format,
 216                        ...) __attribute__((format (printf, 2, 3)));
 217
 218 static NORETURN void bad_object(unsigned long offset, const char *format, ...)
 219 {
 220         va_list params;
 221         char buf[1024];
 222
 223         va_start(params, format);
 224         vsnprintf(buf, sizeof(buf), format, params);
 225         va_end(params);
 226         die(_("pack has bad object at offset %lu: %s"), offset, buf);
 227 }
 228
 229 static struct base_data *alloc_base_data(void)
 230 {
 231         struct base_data *base = xmalloc(sizeof(struct base_data));
 232         memset(base, 0, sizeof(*base));
 233         base->ref_last = -1;
 234         base->ofs_last = -1;
 235         return base;
 236 }
 237
 238 static void free_base_data(struct base_data *c)
 239 {
 240         if (c->data) {
 241                 free(c->data);
 242                 c->data = NULL;
 243                 base_cache_used -= c->size;
 244         }
 245 }
 246
 247 static void prune_base_data(struct base_data *retain)
 248 {
 249         struct base_data *b;
 250         for (b = base_cache;
 251              base_cache_used > delta_base_cache_limit && b;
 252              b = b->child) {
 253                 if (b->data && b != retain)
 254                         free_base_data(b);
 255         }
 256 }
 257
 258 static void link_base_data(struct base_data *base, struct base_data *c)
 259 {
 260         if (base)
 261                 base->child = c;
 262         else
 263                 base_cache = c;
 264
 265         c->base = base;
 266         c->child = NULL;
 267         if (c->data)
 268                 base_cache_used += c->size;
 269         prune_base_data(c);
 270 }
 271
 272 static void unlink_base_data(struct base_data *c)
 273 {
 274         struct base_data *base = c->base;
 275         if (base)
 276                 base->child = NULL;
 277         else
 278                 base_cache = NULL;
 279         free_base_data(c);
 280 }
 281
 282 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 283 {
 284         int status;
 285         git_zstream stream;
 286         void *buf = xmalloc(size);
 287
 288         memset(&stream, 0, sizeof(stream));
 289         git_inflate_init(&stream);
 290         stream.next_out = buf;
 291         stream.avail_out = size;
 292
 293         do {
 294                 stream.next_in = fill(1);
 295                 stream.avail_in = input_len;
 296                 status = git_inflate(&stream, 0);
 297                 use(input_len - stream.avail_in);
 298         } while (status == Z_OK);
 299         if (stream.total_out != size || status != Z_STREAM_END)
 300                 bad_object(offset, _("inflate returned %d"), status);
 301         git_inflate_end(&stream);
 302         return buf;
 303 }
 304
 305 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 306 {
 307         unsigned char *p;
 308         unsigned long size, c;
 309         off_t base_offset;
 310         unsigned shift;
 311         void *data;
 312
 313         obj->idx.offset = consumed_bytes;
 314         input_crc32 = crc32(0, NULL, 0);
 315
 316         p = fill(1);
 317         c = *p;
 318         use(1);
 319         obj->type = (c >> 4) & 7;
 320         size = (c & 15);
 321         shift = 4;
 322         while (c & 0x80) {
 323                 p = fill(1);
 324                 c = *p;
 325                 use(1);
 326                 size += (c & 0x7f) << shift;
 327                 shift += 7;
 328         }
 329         obj->size = size;
 330
 331         switch (obj->type) {
 332         case OBJ_REF_DELTA:
 333                 hashcpy(delta_base->sha1, fill(20));
 334                 use(20);
 335                 break;
 336         case OBJ_OFS_DELTA:
 337                 memset(delta_base, 0, sizeof(*delta_base));
 338                 p = fill(1);
 339                 c = *p;
 340                 use(1);
 341                 base_offset = c & 127;
 342                 while (c & 128) {
 343                         base_offset += 1;
 344                         if (!base_offset || MSB(base_offset, 7))
 345                                 bad_object(obj->idx.offset, _("offset value overflow for delta base object"));
 346                         p = fill(1);
 347                         c = *p;
 348                         use(1);
 349                         base_offset = (base_offset << 7) + (c & 127);
 350                 }
 351                 delta_base->offset = obj->idx.offset - base_offset;
 352                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 353                         bad_object(obj->idx.offset, _("delta base offset is out of bound"));
 354                 break;
 355         case OBJ_COMMIT:
 356         case OBJ_TREE:
 357         case OBJ_BLOB:
 358         case OBJ_TAG:
 359                 break;
 360         default:
 361                 bad_object(obj->idx.offset, _("unknown object type %d"), obj->type);
 362         }
 363         obj->hdr_size = consumed_bytes - obj->idx.offset;
 364
 365         data = unpack_entry_data(obj->idx.offset, obj->size);
 366         obj->idx.crc32 = input_crc32;
 367         return data;
 368 }
 369
 370 static void *get_data_from_pack(struct object_entry *obj)
 371 {
 372         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 373         unsigned long len = obj[1].idx.offset - from;
 374         unsigned char *data, *inbuf;
 375         git_zstream stream;
 376         int status;
 377
 378         data = xmalloc(obj->size);
 379         inbuf = xmalloc((len < 64*1024) ? len : 64*1024);
 380
 381         memset(&stream, 0, sizeof(stream));
 382         git_inflate_init(&stream);
 383         stream.next_out = data;
 384         stream.avail_out = obj->size;
 385
 386         do {
 387                 ssize_t n = (len < 64*1024) ? len : 64*1024;
 388                 n = pread(pack_fd, inbuf, n, from);
 389                 if (n < 0)
 390                         die_errno(_("cannot pread pack file"));
 391                 if (!n)
 392                         die(Q_("premature end of pack file, %lu byte missing",
 393                                "premature end of pack file, %lu bytes missing",
 394                                len),
 395                             len);
 396                 from += n;
 397                 len -= n;
 398                 stream.next_in = inbuf;
 399                 stream.avail_in = n;
 400                 status = git_inflate(&stream, 0);
 401         } while (len && status == Z_OK && !stream.avail_in);
 402
 403         /* This has been inflated OK when first encountered, so... */
 404         if (status != Z_STREAM_END || stream.total_out != obj->size)
 405                 die(_("serious inflate inconsistency"));
 406
 407         git_inflate_end(&stream);
 408         free(inbuf);
 409         return data;
 410 }
 411
 412 static int compare_delta_bases(const union delta_base *base1,
 413                                const union delta_base *base2,
 414                                enum object_type type1,
 415                                enum object_type type2)
 416 {
 417         int cmp = type1 - type2;
 418         if (cmp)
 419                 return cmp;
 420         return memcmp(base1, base2, UNION_BASE_SZ);
 421 }
 422
 423 static int find_delta(const union delta_base *base, enum object_type type)
 424 {
 425         int first = 0, last = nr_deltas;
 426
 427         while (first < last) {
 428                 int next = (first + last) / 2;
 429                 struct delta_entry *delta = &deltas[next];
 430                 int cmp;
 431
 432                 cmp = compare_delta_bases(base, &delta->base,
 433                                           type, objects[delta->obj_no].type);
 434                 if (!cmp)
 435                         return next;
 436                 if (cmp < 0) {
 437                         last = next;
 438                         continue;
 439                 }
 440                 first = next+1;
 441         }
 442         return -first-1;
 443 }
 444
 445 static void find_delta_children(const union delta_base *base,
 446                                 int *first_index, int *last_index,
 447                                 enum object_type type)
 448 {
 449         int first = find_delta(base, type);
 450         int last = first;
 451         int end = nr_deltas - 1;
 452
 453         if (first < 0) {
 454                 *first_index = 0;
 455                 *last_index = -1;
 456                 return;
 457         }
 458         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 459                 --first;
 460         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 461                 ++last;
 462         *first_index = first;
 463         *last_index = last;
 464 }
 465
 466 static void sha1_object(const void *data, unsigned long size,
 467                         enum object_type type, unsigned char *sha1)
 468 {
 469         hash_sha1_file(data, size, typename(type), sha1);
 470         if (has_sha1_file(sha1)) {
 471                 void *has_data;
 472                 enum object_type has_type;
 473                 unsigned long has_size;
 474                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 475                 if (!has_data)
 476                         die(_("cannot read existing object %s"), sha1_to_hex(sha1));
 477                 if (size != has_size || type != has_type ||
 478                     memcmp(data, has_data, size) != 0)
 479                         die(_("SHA1 COLLISION FOUND WITH %s !"), sha1_to_hex(sha1));
 480                 free(has_data);
 481         }
 482         if (strict) {
 483                 if (type == OBJ_BLOB) {
 484                         struct blob *blob = lookup_blob(sha1);
 485                         if (blob)
 486                                 blob->object.flags |= FLAG_CHECKED;
 487                         else
 488                                 die(_("invalid blob object %s"), sha1_to_hex(sha1));
 489                 } else {
 490                         struct object *obj;
 491                         int eaten;
 492                         void *buf = (void *) data;
 493
 494                         /*
 495                          * we do not need to free the memory here, as the
 496                          * buf is deleted by the caller.
 497                          */
 498                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 499                         if (!obj)
 500                                 die(_("invalid %s"), typename(type));
 501                         if (fsck_object(obj, 1, fsck_error_function))
 502                                 die(_("Error in object"));
 503                         if (fsck_walk(obj, mark_link, NULL))
 504                                 die(_("Not all child objects of %s are reachable"), sha1_to_hex(obj->sha1));
 505
 506                         if (obj->type == OBJ_TREE) {
 507                                 struct tree *item = (struct tree *) obj;
 508                                 item->buffer = NULL;
 509                         }
 510                         if (obj->type == OBJ_COMMIT) {
 511                                 struct commit *commit = (struct commit *) obj;
 512                                 commit->buffer = NULL;
 513                         }
 514                         obj->flags |= FLAG_CHECKED;
 515                 }
 516         }
 517 }
 518
 519 static int is_delta_type(enum object_type type)
 520 {
 521         return (type == OBJ_REF_DELTA || type == OBJ_OFS_DELTA);
 522 }
 523
 524 /*
 525  * This function is part of find_unresolved_deltas(). There are two
 526  * walkers going in the opposite ways.
 527  *
 528  * The first one in find_unresolved_deltas() traverses down from
 529  * parent node to children, deflating nodes along the way. However,
 530  * memory for deflated nodes is limited by delta_base_cache_limit, so
 531  * at some point parent node's deflated content may be freed.
 532  *
 533  * The second walker is this function, which goes from current node up
 534  * to top parent if necessary to deflate the node. In normal
 535  * situation, its parent node would be already deflated, so it just
 536  * needs to apply delta.
 537  *
 538  * In the worst case scenario, parent node is no longer deflated because
 539  * we're running out of delta_base_cache_limit; we need to re-deflate
 540  * parents, possibly up to the top base.
 541  *
 542  * All deflated objects here are subject to be freed if we exceed
 543  * delta_base_cache_limit, just like in find_unresolved_deltas(), we
 544  * just need to make sure the last node is not freed.
 545  */
 546 static void *get_base_data(struct base_data *c)
 547 {
 548         if (!c->data) {
 549                 struct object_entry *obj = c->obj;
 550                 struct base_data **delta = NULL;
 551                 int delta_nr = 0, delta_alloc = 0;
 552
 553                 while (is_delta_type(c->obj->type) && !c->data) {
 554                         ALLOC_GROW(delta, delta_nr + 1, delta_alloc);
 555                         delta[delta_nr++] = c;
 556                         c = c->base;
 557                 }
 558                 if (!delta_nr) {
 559                         c->data = get_data_from_pack(obj);
 560                         c->size = obj->size;
 561                         base_cache_used += c->size;
 562                         prune_base_data(c);
 563                 }
 564                 for (; delta_nr > 0; delta_nr--) {
 565                         void *base, *raw;
 566                         c = delta[delta_nr - 1];
 567                         obj = c->obj;
 568                         base = get_base_data(c->base);
 569                         raw = get_data_from_pack(obj);
 570                         c->data = patch_delta(
 571                                 base, c->base->size,
 572                                 raw, obj->size,
 573                                 &c->size);
 574                         free(raw);
 575                         if (!c->data)
 576                                 bad_object(obj->idx.offset, _("failed to apply delta"));
 577                         base_cache_used += c->size;
 578                         prune_base_data(c);
 579                 }
 580                 free(delta);
 581         }
 582         return c->data;
 583 }
 584
 585 static void resolve_delta(struct object_entry *delta_obj,
 586                           struct base_data *base, struct base_data *result)
 587 {
 588         void *base_data, *delta_data;
 589
 590         delta_obj->real_type = base->obj->real_type;
 591         delta_obj->delta_depth = base->obj->delta_depth + 1;
 592         if (deepest_delta < delta_obj->delta_depth)
 593                 deepest_delta = delta_obj->delta_depth;
 594         delta_obj->base_object_no = base->obj - objects;
 595         delta_data = get_data_from_pack(delta_obj);
 596         base_data = get_base_data(base);
 597         result->obj = delta_obj;
 598         result->data = patch_delta(base_data, base->size,
 599                                    delta_data, delta_obj->size, &result->size);
 600         free(delta_data);
 601         if (!result->data)
 602                 bad_object(delta_obj->idx.offset, _("failed to apply delta"));
 603         sha1_object(result->data, result->size, delta_obj->real_type,
 604                     delta_obj->idx.sha1);
 605         nr_resolved_deltas++;
 606 }
 607
 608 static struct base_data *find_unresolved_deltas_1(struct base_data *base,
 609                                                   struct base_data *prev_base)
 610 {
 611         if (base->ref_last == -1 && base->ofs_last == -1) {
 612                 union delta_base base_spec;
 613
 614                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 615                 find_delta_children(&base_spec,
 616                                     &base->ref_first, &base->ref_last, OBJ_REF_DELTA);
 617
 618                 memset(&base_spec, 0, sizeof(base_spec));
 619                 base_spec.offset = base->obj->idx.offset;
 620                 find_delta_children(&base_spec,
 621                                     &base->ofs_first, &base->ofs_last, OBJ_OFS_DELTA);
 622
 623                 if (base->ref_last == -1 && base->ofs_last == -1) {
 624                         free(base->data);
 625                         return NULL;
 626                 }
 627
 628                 link_base_data(prev_base, base);
 629         }
 630
 631         if (base->ref_first <= base->ref_last) {
 632                 struct object_entry *child = objects + deltas[base->ref_first].obj_no;
 633                 struct base_data *result = alloc_base_data();
 634
 635                 assert(child->real_type == OBJ_REF_DELTA);
 636                 resolve_delta(child, base, result);
 637                 if (base->ref_first == base->ref_last && base->ofs_last == -1)
 638                         free_base_data(base);
 639
 640                 base->ref_first++;
 641                 return result;
 642         }
 643
 644         if (base->ofs_first <= base->ofs_last) {
 645                 struct object_entry *child = objects + deltas[base->ofs_first].obj_no;
 646                 struct base_data *result = alloc_base_data();
 647
 648                 assert(child->real_type == OBJ_OFS_DELTA);
 649                 resolve_delta(child, base, result);
 650                 if (base->ofs_first == base->ofs_last)
 651                         free_base_data(base);
 652
 653                 base->ofs_first++;
 654                 return result;
 655         }
 656
 657         unlink_base_data(base);
 658         return NULL;
 659 }
 660
 661 static void find_unresolved_deltas(struct base_data *base)
 662 {
 663         struct base_data *new_base, *prev_base = NULL;
 664         for (;;) {
 665                 new_base = find_unresolved_deltas_1(base, prev_base);
 666
 667                 if (new_base) {
 668                         prev_base = base;
 669                         base = new_base;
 670                 } else {
 671                         free(base);
 672                         base = prev_base;
 673                         if (!base)
 674                                 return;
 675                         prev_base = base->base;
 676                 }
 677         }
 678 }
 679
 680 static int compare_delta_entry(const void *a, const void *b)
 681 {
 682         const struct delta_entry *delta_a = a;
 683         const struct delta_entry *delta_b = b;
 684
 685         /* group by type (ref vs ofs) and then by value (sha-1 or offset) */
 686         return compare_delta_bases(&delta_a->base, &delta_b->base,
 687                                    objects[delta_a->obj_no].type,
 688                                    objects[delta_b->obj_no].type);
 689 }
 690
 691 /* Parse all objects and return the pack content SHA1 hash */
 692 static void parse_pack_objects(unsigned char *sha1)
 693 {
 694         int i;
 695         struct delta_entry *delta = deltas;
 696         struct stat st;
 697
 698         /*
 699          * First pass:
 700          * - find locations of all objects;
 701          * - calculate SHA1 of all non-delta objects;
 702          * - remember base (SHA1 or offset) for all deltas.
 703          */
 704         if (verbose)
 705                 progress = start_progress(
 706                                 from_stdin ? _("Receiving objects") : _("Indexing objects"),
 707                                 nr_objects);
 708         for (i = 0; i < nr_objects; i++) {
 709                 struct object_entry *obj = &objects[i];
 710                 void *data = unpack_raw_entry(obj, &delta->base);
 711                 obj->real_type = obj->type;
 712                 if (is_delta_type(obj->type)) {
 713                         nr_deltas++;
 714                         delta->obj_no = i;
 715                         delta++;
 716                 } else
 717                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 718                 free(data);
 719                 display_progress(progress, i+1);
 720         }
 721         objects[i].idx.offset = consumed_bytes;
 722         stop_progress(&progress);
 723
 724         /* Check pack integrity */
 725         flush();
 726         git_SHA1_Final(sha1, &input_ctx);
 727         if (hashcmp(fill(20), sha1))
 728                 die(_("pack is corrupted (SHA1 mismatch)"));
 729         use(20);
 730
 731         /* If input_fd is a file, we should have reached its end now. */
 732         if (fstat(input_fd, &st))
 733                 die_errno(_("cannot fstat packfile"));
 734         if (S_ISREG(st.st_mode) &&
 735                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 736                 die(_("pack has junk at the end"));
 737
 738         if (!nr_deltas)
 739                 return;
 740
 741         /* Sort deltas by base SHA1/offset for fast searching */
 742         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 743               compare_delta_entry);
 744
 745         /*
 746          * Second pass:
 747          * - for all non-delta objects, look if it is used as a base for
 748          *   deltas;
 749          * - if used as a base, uncompress the object and apply all deltas,
 750          *   recursively checking if the resulting object is used as a base
 751          *   for some more deltas.
 752          */
 753         if (verbose)
 754                 progress = start_progress(_("Resolving deltas"), nr_deltas);
 755         for (i = 0; i < nr_objects; i++) {
 756                 struct object_entry *obj = &objects[i];
 757                 struct base_data *base_obj = alloc_base_data();
 758
 759                 if (is_delta_type(obj->type))
 760                         continue;
 761                 base_obj->obj = obj;
 762                 base_obj->data = NULL;
 763                 find_unresolved_deltas(base_obj);
 764                 display_progress(progress, nr_resolved_deltas);
 765         }
 766 }
 767
 768 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 769 {
 770         git_zstream stream;
 771         int status;
 772         unsigned char outbuf[4096];
 773
 774         memset(&stream, 0, sizeof(stream));
 775         git_deflate_init(&stream, zlib_compression_level);
 776         stream.next_in = in;
 777         stream.avail_in = size;
 778
 779         do {
 780                 stream.next_out = outbuf;
 781                 stream.avail_out = sizeof(outbuf);
 782                 status = git_deflate(&stream, Z_FINISH);
 783                 sha1write(f, outbuf, sizeof(outbuf) - stream.avail_out);
 784         } while (status == Z_OK);
 785
 786         if (status != Z_STREAM_END)
 787                 die(_("unable to deflate appended object (%d)"), status);
 788         size = stream.total_out;
 789         git_deflate_end(&stream);
 790         return size;
 791 }
 792
 793 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 794                                const unsigned char *sha1, void *buf,
 795                                unsigned long size, enum object_type type)
 796 {
 797         struct object_entry *obj = &objects[nr_objects++];
 798         unsigned char header[10];
 799         unsigned long s = size;
 800         int n = 0;
 801         unsigned char c = (type << 4) | (s & 15);
 802         s >>= 4;
 803         while (s) {
 804                 header[n++] = c | 0x80;
 805                 c = s & 0x7f;
 806                 s >>= 7;
 807         }
 808         header[n++] = c;
 809         crc32_begin(f);
 810         sha1write(f, header, n);
 811         obj[0].size = size;
 812         obj[0].hdr_size = n;
 813         obj[0].type = type;
 814         obj[0].real_type = type;
 815         obj[1].idx.offset = obj[0].idx.offset + n;
 816         obj[1].idx.offset += write_compressed(f, buf, size);
 817         obj[0].idx.crc32 = crc32_end(f);
 818         sha1flush(f);
 819         hashcpy(obj->idx.sha1, sha1);
 820         return obj;
 821 }
 822
 823 static int delta_pos_compare(const void *_a, const void *_b)
 824 {
 825         struct delta_entry *a = *(struct delta_entry **)_a;
 826         struct delta_entry *b = *(struct delta_entry **)_b;
 827         return a->obj_no - b->obj_no;
 828 }
 829
 830 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 831 {
 832         struct delta_entry **sorted_by_pos;
 833         int i, n = 0;
 834
 835         /*
 836          * Since many unresolved deltas may well be themselves base objects
 837          * for more unresolved deltas, we really want to include the
 838          * smallest number of base objects that would cover as much delta
 839          * as possible by picking the
 840          * trunc deltas first, allowing for other deltas to resolve without
 841          * additional base objects.  Since most base objects are to be found
 842          * before deltas depending on them, a good heuristic is to start
 843          * resolving deltas in the same order as their position in the pack.
 844          */
 845         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 846         for (i = 0; i < nr_deltas; i++) {
 847                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 848                         continue;
 849                 sorted_by_pos[n++] = &deltas[i];
 850         }
 851         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 852
 853         for (i = 0; i < n; i++) {
 854                 struct delta_entry *d = sorted_by_pos[i];
 855                 enum object_type type;
 856                 struct base_data *base_obj = alloc_base_data();
 857
 858                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 859                         continue;
 860                 base_obj->data = read_sha1_file(d->base.sha1, &type, &base_obj->size);
 861                 if (!base_obj->data)
 862                         continue;
 863
 864                 if (check_sha1_signature(d->base.sha1, base_obj->data,
 865                                 base_obj->size, typename(type)))
 866                         die(_("local object %s is corrupt"), sha1_to_hex(d->base.sha1));
 867                 base_obj->obj = append_obj_to_pack(f, d->base.sha1,
 868                                         base_obj->data, base_obj->size, type);
 869                 find_unresolved_deltas(base_obj);
 870                 display_progress(progress, nr_resolved_deltas);
 871         }
 872         free(sorted_by_pos);
 873 }
 874
 875 static void final(const char *final_pack_name, const char *curr_pack_name,
 876                   const char *final_index_name, const char *curr_index_name,
 877                   const char *keep_name, const char *keep_msg,
 878                   unsigned char *sha1)
 879 {
 880         const char *report = "pack";
 881         char name[PATH_MAX];
 882         int err;
 883
 884         if (!from_stdin) {
 885                 close(input_fd);
 886         } else {
 887                 fsync_or_die(output_fd, curr_pack_name);
 888                 err = close(output_fd);
 889                 if (err)
 890                         die_errno(_("error while closing pack file"));
 891         }
 892
 893         if (keep_msg) {
 894                 int keep_fd, keep_msg_len = strlen(keep_msg);
 895
 896                 if (!keep_name)
 897                         keep_fd = odb_pack_keep(name, sizeof(name), sha1);
 898                 else
 899                         keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 900
 901                 if (keep_fd < 0) {
 902                         if (errno != EEXIST)
 903                                 die_errno(_("cannot write keep file '%s'"),
 904                                           keep_name);
 905                 } else {
 906                         if (keep_msg_len > 0) {
 907                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 908                                 write_or_die(keep_fd, "\n", 1);
 909                         }
 910                         if (close(keep_fd) != 0)
 911                                 die_errno(_("cannot close written keep file '%s'"),
 912                                     keep_name);
 913                         report = "keep";
 914                 }
 915         }
 916
 917         if (final_pack_name != curr_pack_name) {
 918                 if (!final_pack_name) {
 919                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 920                                  get_object_directory(), sha1_to_hex(sha1));
 921                         final_pack_name = name;
 922                 }
 923                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 924                         die(_("cannot store pack file"));
 925         } else if (from_stdin)
 926                 chmod(final_pack_name, 0444);
 927
 928         if (final_index_name != curr_index_name) {
 929                 if (!final_index_name) {
 930                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 931                                  get_object_directory(), sha1_to_hex(sha1));
 932                         final_index_name = name;
 933                 }
 934                 if (move_temp_to_file(curr_index_name, final_index_name))
 935                         die(_("cannot store index file"));
 936         } else
 937                 chmod(final_index_name, 0444);
 938
 939         if (!from_stdin) {
 940                 printf("%s\n", sha1_to_hex(sha1));
 941         } else {
 942                 char buf[48];
 943                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 944                                    report, sha1_to_hex(sha1));
 945                 write_or_die(1, buf, len);
 946
 947                 /*
 948                  * Let's just mimic git-unpack-objects here and write
 949                  * the last part of the input buffer to stdout.
 950                  */
 951                 while (input_len) {
 952                         err = xwrite(1, input_buffer + input_offset, input_len);
 953                         if (err <= 0)
 954                                 break;
 955                         input_len -= err;
 956                         input_offset += err;
 957                 }
 958         }
 959 }
 960
 961 static int git_index_pack_config(const char *k, const char *v, void *cb)
 962 {
 963         struct pack_idx_option *opts = cb;
 964
 965         if (!strcmp(k, "pack.indexversion")) {
 966                 opts->version = git_config_int(k, v);
 967                 if (opts->version > 2)
 968                         die("bad pack.indexversion=%"PRIu32, opts->version);
 969                 return 0;
 970         }
 971         return git_default_config(k, v, cb);
 972 }
 973
 974 static int cmp_uint32(const void *a_, const void *b_)
 975 {
 976         uint32_t a = *((uint32_t *)a_);
 977         uint32_t b = *((uint32_t *)b_);
 978
 979         return (a < b) ? -1 : (a != b);
 980 }
 981
 982 static void read_v2_anomalous_offsets(struct packed_git *p,
 983                                       struct pack_idx_option *opts)
 984 {
 985         const uint32_t *idx1, *idx2;
 986         uint32_t i;
 987
 988         /* The address of the 4-byte offset table */
 989         idx1 = (((const uint32_t *)p->index_data)
 990                 + 2 /* 8-byte header */
 991                 + 256 /* fan out */
 992                 + 5 * p->num_objects /* 20-byte SHA-1 table */
 993                 + p->num_objects /* CRC32 table */
 994                 );
 995
 996         /* The address of the 8-byte offset table */
 997         idx2 = idx1 + p->num_objects;
 998
 999         for (i = 0; i < p->num_objects; i++) {
1000                 uint32_t off = ntohl(idx1[i]);
1001                 if (!(off & 0x80000000))
1002                         continue;
1003                 off = off & 0x7fffffff;
1004                 if (idx2[off * 2])
1005                         continue;
1006                 /*
1007                  * The real offset is ntohl(idx2[off * 2]) in high 4
1008                  * octets, and ntohl(idx2[off * 2 + 1]) in low 4
1009                  * octets.  But idx2[off * 2] is Zero!!!
1010                  */
1011                 ALLOC_GROW(opts->anomaly, opts->anomaly_nr + 1, opts->anomaly_alloc);
1012                 opts->anomaly[opts->anomaly_nr++] = ntohl(idx2[off * 2 + 1]);
1013         }
1014
1015         if (1 < opts->anomaly_nr)
1016                 qsort(opts->anomaly, opts->anomaly_nr, sizeof(uint32_t), cmp_uint32);
1017 }
1018
1019 static void read_idx_option(struct pack_idx_option *opts, const char *pack_name)
1020 {
1021         struct packed_git *p = add_packed_git(pack_name, strlen(pack_name), 1);
1022
1023         if (!p)
1024                 die(_("Cannot open existing pack file '%s'"), pack_name);
1025         if (open_pack_index(p))
1026                 die(_("Cannot open existing pack idx file for '%s'"), pack_name);
1027
1028         /* Read the attributes from the existing idx file */
1029         opts->version = p->index_version;
1030
1031         if (opts->version == 2)
1032                 read_v2_anomalous_offsets(p, opts);
1033
1034         /*
1035          * Get rid of the idx file as we do not need it anymore.
1036          * NEEDSWORK: extract this bit from free_pack_by_name() in
1037          * sha1_file.c, perhaps?  It shouldn't matter very much as we
1038          * know we haven't installed this pack (hence we never have
1039          * read anything from it).
1040          */
1041         close_pack_index(p);
1042         free(p);
1043 }
1044
1045 static void show_pack_info(int stat_only)
1046 {
1047         int i, baseobjects = nr_objects - nr_deltas;
1048         unsigned long *chain_histogram = NULL;
1049
1050         if (deepest_delta)
1051                 chain_histogram = xcalloc(deepest_delta, sizeof(unsigned long));
1052
1053         for (i = 0; i < nr_objects; i++) {
1054                 struct object_entry *obj = &objects[i];
1055
1056                 if (is_delta_type(obj->type))
1057                         chain_histogram[obj->delta_depth - 1]++;
1058                 if (stat_only)
1059                         continue;
1060                 printf("%s %-6s %lu %lu %"PRIuMAX,
1061                        sha1_to_hex(obj->idx.sha1),
1062                        typename(obj->real_type), obj->size,
1063                        (unsigned long)(obj[1].idx.offset - obj->idx.offset),
1064                        (uintmax_t)obj->idx.offset);
1065                 if (is_delta_type(obj->type)) {
1066                         struct object_entry *bobj = &objects[obj->base_object_no];
1067                         printf(" %u %s", obj->delta_depth, sha1_to_hex(bobj->idx.sha1));
1068                 }
1069                 putchar('\n');
1070         }
1071
1072         if (baseobjects)
1073                 printf_ln(Q_("non delta: %d object",
1074                              "non delta: %d objects",
1075                              baseobjects),
1076                           baseobjects);
1077         for (i = 0; i < deepest_delta; i++) {
1078                 if (!chain_histogram[i])
1079                         continue;
1080                 printf_ln(Q_("chain length = %d: %lu object",
1081                              "chain length = %d: %lu objects",
1082                              chain_histogram[i]),
1083                           i + 1,
1084                           chain_histogram[i]);
1085         }
1086 }
1087
1088 int cmd_index_pack(int argc, const char **argv, const char *prefix)
1089 {
1090         int i, fix_thin_pack = 0, verify = 0, stat_only = 0, stat = 0;
1091         const char *curr_pack, *curr_index;
1092         const char *index_name = NULL, *pack_name = NULL;
1093         const char *keep_name = NULL, *keep_msg = NULL;
1094         char *index_name_buf = NULL, *keep_name_buf = NULL;
1095         struct pack_idx_entry **idx_objects;
1096         struct pack_idx_option opts;
1097         unsigned char pack_sha1[20];
1098
1099         if (argc == 2 && !strcmp(argv[1], "-h"))
1100                 usage(index_pack_usage);
1101
1102         read_replace_refs = 0;
1103
1104         reset_pack_idx_option(&opts);
1105         git_config(git_index_pack_config, &opts);
1106         if (prefix && chdir(prefix))
1107                 die(_("Cannot come back to cwd"));
1108
1109         for (i = 1; i < argc; i++) {
1110                 const char *arg = argv[i];
1111
1112                 if (*arg == '-') {
1113                         if (!strcmp(arg, "--stdin")) {
1114                                 from_stdin = 1;
1115                         } else if (!strcmp(arg, "--fix-thin")) {
1116                                 fix_thin_pack = 1;
1117                         } else if (!strcmp(arg, "--strict")) {
1118                                 strict = 1;
1119                         } else if (!strcmp(arg, "--verify")) {
1120                                 verify = 1;
1121                         } else if (!strcmp(arg, "--verify-stat")) {
1122                                 verify = 1;
1123                                 stat = 1;
1124                         } else if (!strcmp(arg, "--verify-stat-only")) {
1125                                 verify = 1;
1126                                 stat = 1;
1127                                 stat_only = 1;
1128                         } else if (!strcmp(arg, "--keep")) {
1129                                 keep_msg = "";
1130                         } else if (!prefixcmp(arg, "--keep=")) {
1131                                 keep_msg = arg + 7;
1132                         } else if (!prefixcmp(arg, "--pack_header=")) {
1133                                 struct pack_header *hdr;
1134                                 char *c;
1135
1136                                 hdr = (struct pack_header *)input_buffer;
1137                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
1138                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
1139                                 if (*c != ',')
1140                                         die(_("bad %s"), arg);
1141                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
1142                                 if (*c)
1143                                         die(_("bad %s"), arg);
1144                                 input_len = sizeof(*hdr);
1145                         } else if (!strcmp(arg, "-v")) {
1146                                 verbose = 1;
1147                         } else if (!strcmp(arg, "-o")) {
1148                                 if (index_name || (i+1) >= argc)
1149                                         usage(index_pack_usage);
1150                                 index_name = argv[++i];
1151                         } else if (!prefixcmp(arg, "--index-version=")) {
1152                                 char *c;
1153                                 opts.version = strtoul(arg + 16, &c, 10);
1154                                 if (opts.version > 2)
1155                                         die(_("bad %s"), arg);
1156                                 if (*c == ',')
1157                                         opts.off32_limit = strtoul(c+1, &c, 0);
1158                                 if (*c || opts.off32_limit & 0x80000000)
1159                                         die(_("bad %s"), arg);
1160                         } else
1161                                 usage(index_pack_usage);
1162                         continue;
1163                 }
1164
1165                 if (pack_name)
1166                         usage(index_pack_usage);
1167                 pack_name = arg;
1168         }
1169
1170         if (!pack_name && !from_stdin)
1171                 usage(index_pack_usage);
1172         if (fix_thin_pack && !from_stdin)
1173                 die(_("--fix-thin cannot be used without --stdin"));
1174         if (!index_name && pack_name) {
1175                 int len = strlen(pack_name);
1176                 if (!has_extension(pack_name, ".pack"))
1177                         die(_("packfile name '%s' does not end with '.pack'"),
1178                             pack_name);
1179                 index_name_buf = xmalloc(len);
1180                 memcpy(index_name_buf, pack_name, len - 5);
1181                 strcpy(index_name_buf + len - 5, ".idx");
1182                 index_name = index_name_buf;
1183         }
1184         if (keep_msg && !keep_name && pack_name) {
1185                 int len = strlen(pack_name);
1186                 if (!has_extension(pack_name, ".pack"))
1187                         die(_("packfile name '%s' does not end with '.pack'"),
1188                             pack_name);
1189                 keep_name_buf = xmalloc(len);
1190                 memcpy(keep_name_buf, pack_name, len - 5);
1191                 strcpy(keep_name_buf + len - 5, ".keep");
1192                 keep_name = keep_name_buf;
1193         }
1194         if (verify) {
1195                 if (!index_name)
1196                         die(_("--verify with no packfile name given"));
1197                 read_idx_option(&opts, index_name);
1198                 opts.flags |= WRITE_IDX_VERIFY | WRITE_IDX_STRICT;
1199         }
1200         if (strict)
1201                 opts.flags |= WRITE_IDX_STRICT;
1202
1203         curr_pack = open_pack_file(pack_name);
1204         parse_pack_header();
1205         objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
1206         deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
1207         parse_pack_objects(pack_sha1);
1208         if (nr_deltas == nr_resolved_deltas) {
1209                 stop_progress(&progress);
1210                 /* Flush remaining pack final 20-byte SHA1. */
1211                 flush();
1212         } else {
1213                 if (fix_thin_pack) {
1214                         struct sha1file *f;
1215                         unsigned char read_sha1[20], tail_sha1[20];
1216                         char msg[48];
1217                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
1218                         int nr_objects_initial = nr_objects;
1219                         if (nr_unresolved <= 0)
1220                                 die(_("confusion beyond insanity"));
1221                         objects = xrealloc(objects,
1222                                            (nr_objects + nr_unresolved + 1)
1223                                            * sizeof(*objects));
1224                         f = sha1fd(output_fd, curr_pack);
1225                         fix_unresolved_deltas(f, nr_unresolved);
1226                         sprintf(msg, "completed with %d local objects",
1227                                 nr_objects - nr_objects_initial);
1228                         stop_progress_msg(&progress, msg);
1229                         sha1close(f, tail_sha1, 0);
1230                         hashcpy(read_sha1, pack_sha1);
1231                         fixup_pack_header_footer(output_fd, pack_sha1,
1232                                                  curr_pack, nr_objects,
1233                                                  read_sha1, consumed_bytes-20);
1234                         if (hashcmp(read_sha1, tail_sha1) != 0)
1235                                 die("Unexpected tail checksum for %s "
1236                                     "(disk corruption?)", curr_pack);
1237                 }
1238                 if (nr_deltas != nr_resolved_deltas)
1239                         die(Q_("pack has %d unresolved delta",
1240                                "pack has %d unresolved deltas",
1241                                nr_deltas - nr_resolved_deltas),
1242                             nr_deltas - nr_resolved_deltas);
1243         }
1244         free(deltas);
1245         if (strict)
1246                 check_objects();
1247
1248         if (stat)
1249                 show_pack_info(stat_only);
1250
1251         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1252         for (i = 0; i < nr_objects; i++)
1253                 idx_objects[i] = &objects[i].idx;
1254         curr_index = write_idx_file(index_name, idx_objects, nr_objects, &opts, pack_sha1);
1255         free(idx_objects);
1256
1257         if (!verify)
1258                 final(pack_name, curr_pack,
1259                       index_name, curr_index,
1260                       keep_name, keep_msg,
1261                       pack_sha1);
1262         else
1263                 close(input_fd);
1264         free(objects);
1265         free(index_name_buf);
1266         free(keep_name_buf);
1267         if (pack_name == NULL)
1268                 free((void *) curr_pack);
1269         if (index_name == NULL)
1270                 free((void *) curr_index);
1271
1272         return 0;
1273 }