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