src/indexer.c

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #include "git2/indexer.h"
   9 #include "git2/object.h"
  10
  11 #include "common.h"
  12 #include "pack.h"
  13 #include "mwindow.h"
  14 #include "posix.h"
  15 #include "pack.h"
  16 #include "filebuf.h"
  17 #include "oid.h"
  18 #include "oidmap.h"
  19 #include "zstream.h"
  20
  21 GIT__USE_OIDMAP
  22
  23 extern git_mutex git__mwindow_mutex;
  24
  25 #define UINT31_MAX (0x7FFFFFFF)
  26
  27 struct entry {
  28         git_oid oid;
  29         uint32_t crc;
  30         uint32_t offset;
  31         uint64_t offset_long;
  32 };
  33
  34 struct git_indexer {
  35         unsigned int parsed_header :1,
  36                 opened_pack :1,
  37                 have_stream :1,
  38                 have_delta :1;
  39         struct git_pack_header hdr;
  40         struct git_pack_file *pack;
  41         unsigned int mode;
  42         git_off_t off;
  43         git_off_t entry_start;
  44         git_packfile_stream stream;
  45         size_t nr_objects;
  46         git_vector objects;
  47         git_vector deltas;
  48         unsigned int fanout[256];
  49         git_hash_ctx hash_ctx;
  50         git_oid hash;
  51         git_transfer_progress_cb progress_cb;
  52         void *progress_payload;
  53         char objbuf[8*1024];
  54
  55         /* Needed to look up objects which we want to inject to fix a thin pack */
  56         git_odb *odb;
  57
  58         /* Fields for calculating the packfile trailer (hash of everything before it) */
  59         char inbuf[GIT_OID_RAWSZ];
  60         size_t inbuf_len;
  61         git_hash_ctx trailer;
  62 };
  63
  64 struct delta_info {
  65         git_off_t delta_off;
  66 };
  67
  68 const git_oid *git_indexer_hash(const git_indexer *idx)
  69 {
  70         return &idx->hash;
  71 }
  72
  73 static int parse_header(struct git_pack_header *hdr, struct git_pack_file *pack)
  74 {
  75         int error;
  76         git_map map;
  77
  78         if ((error = p_mmap(&map, sizeof(*hdr), GIT_PROT_READ, GIT_MAP_SHARED, pack->mwf.fd, 0)) < 0)
  79                 return error;
  80
  81         memcpy(hdr, map.data, sizeof(*hdr));
  82         p_munmap(&map);
  83
  84         /* Verify we recognize this pack file format. */
  85         if (hdr->hdr_signature != ntohl(PACK_SIGNATURE)) {
  86                 giterr_set(GITERR_INDEXER, "Wrong pack signature");
  87                 return -1;
  88         }
  89
  90         if (!pack_version_ok(hdr->hdr_version)) {
  91                 giterr_set(GITERR_INDEXER, "Wrong pack version");
  92                 return -1;
  93         }
  94
  95         return 0;
  96 }
  97
  98 static int objects_cmp(const void *a, const void *b)
  99 {
 100         const struct entry *entrya = a;
 101         const struct entry *entryb = b;
 102
 103         return git_oid__cmp(&entrya->oid, &entryb->oid);
 104 }
 105
 106 int git_indexer_new(
 107                 git_indexer **out,
 108                 const char *prefix,
 109                 unsigned int mode,
 110                 git_odb *odb,
 111                 git_transfer_progress_cb progress_cb,
 112                 void *progress_payload)
 113 {
 114         git_indexer *idx;
 115         git_buf path = GIT_BUF_INIT, tmp_path = GIT_BUF_INIT;
 116         static const char suff[] = "/pack";
 117         int error, fd = -1;
 118
 119         idx = git__calloc(1, sizeof(git_indexer));
 120         GITERR_CHECK_ALLOC(idx);
 121         idx->odb = odb;
 122         idx->progress_cb = progress_cb;
 123         idx->progress_payload = progress_payload;
 124         idx->mode = mode ? mode : GIT_PACK_FILE_MODE;
 125         git_hash_ctx_init(&idx->hash_ctx);
 126         git_hash_ctx_init(&idx->trailer);
 127
 128         error = git_buf_joinpath(&path, prefix, suff);
 129         if (error < 0)
 130                 goto cleanup;
 131
 132         fd = git_futils_mktmp(&tmp_path, git_buf_cstr(&path), idx->mode);
 133         git_buf_free(&path);
 134         if (fd < 0)
 135                 goto cleanup;
 136
 137         error = git_packfile_alloc(&idx->pack, git_buf_cstr(&tmp_path));
 138         git_buf_free(&tmp_path);
 139
 140         if (error < 0)
 141                 goto cleanup;
 142
 143         idx->pack->mwf.fd = fd;
 144         if ((error = git_mwindow_file_register(&idx->pack->mwf)) < 0)
 145                 goto cleanup;
 146
 147         *out = idx;
 148         return 0;
 149
 150 cleanup:
 151         if (fd != -1)
 152                 p_close(fd);
 153
 154         git_buf_free(&path);
 155         git_buf_free(&tmp_path);
 156         git__free(idx);
 157         return -1;
 158 }
 159
 160 /* Try to store the delta so we can try to resolve it later */
 161 static int store_delta(git_indexer *idx)
 162 {
 163         struct delta_info *delta;
 164
 165         delta = git__calloc(1, sizeof(struct delta_info));
 166         GITERR_CHECK_ALLOC(delta);
 167         delta->delta_off = idx->entry_start;
 168
 169         if (git_vector_insert(&idx->deltas, delta) < 0)
 170                 return -1;
 171
 172         return 0;
 173 }
 174
 175 static void hash_header(git_hash_ctx *ctx, git_off_t len, git_otype type)
 176 {
 177         char buffer[64];
 178         size_t hdrlen;
 179
 180         hdrlen = git_odb__format_object_header(buffer, sizeof(buffer), (size_t)len, type);
 181         git_hash_update(ctx, buffer, hdrlen);
 182 }
 183
 184 static int hash_object_stream(git_indexer*idx, git_packfile_stream *stream)
 185 {
 186         ssize_t read;
 187
 188         assert(idx && stream);
 189
 190         do {
 191                 if ((read = git_packfile_stream_read(stream, idx->objbuf, sizeof(idx->objbuf))) < 0)
 192                         break;
 193
 194                 git_hash_update(&idx->hash_ctx, idx->objbuf, read);
 195         } while (read > 0);
 196
 197         if (read < 0)
 198                 return (int)read;
 199
 200         return 0;
 201 }
 202
 203 /* In order to create the packfile stream, we need to skip over the delta base description */
 204 static int advance_delta_offset(git_indexer *idx, git_otype type)
 205 {
 206         git_mwindow *w = NULL;
 207
 208         assert(type == GIT_OBJ_REF_DELTA || type == GIT_OBJ_OFS_DELTA);
 209
 210         if (type == GIT_OBJ_REF_DELTA) {
 211                 idx->off += GIT_OID_RAWSZ;
 212         } else {
 213                 git_off_t base_off = get_delta_base(idx->pack, &w, &idx->off, type, idx->entry_start);
 214                 git_mwindow_close(&w);
 215                 if (base_off < 0)
 216                         return (int)base_off;
 217         }
 218
 219         return 0;
 220 }
 221
 222 /* Read from the stream and discard any output */
 223 static int read_object_stream(git_indexer *idx, git_packfile_stream *stream)
 224 {
 225         ssize_t read;
 226
 227         assert(stream);
 228
 229         do {
 230                 read = git_packfile_stream_read(stream, idx->objbuf, sizeof(idx->objbuf));
 231         } while (read > 0);
 232
 233         if (read < 0)
 234                 return (int)read;
 235
 236         return 0;
 237 }
 238
 239 static int crc_object(uint32_t *crc_out, git_mwindow_file *mwf, git_off_t start, git_off_t size)
 240 {
 241         void *ptr;
 242         uint32_t crc;
 243         unsigned int left, len;
 244         git_mwindow *w = NULL;
 245
 246         crc = crc32(0L, Z_NULL, 0);
 247         while (size) {
 248                 ptr = git_mwindow_open(mwf, &w, start, (size_t)size, &left);
 249                 if (ptr == NULL)
 250                         return -1;
 251
 252                 len = min(left, (unsigned int)size);
 253                 crc = crc32(crc, ptr, len);
 254                 size -= len;
 255                 start += len;
 256                 git_mwindow_close(&w);
 257         }
 258
 259         *crc_out = htonl(crc);
 260         return 0;
 261 }
 262
 263 static int store_object(git_indexer *idx)
 264 {
 265         int i, error;
 266         khiter_t k;
 267         git_oid oid;
 268         struct entry *entry;
 269         git_off_t entry_size;
 270         struct git_pack_entry *pentry;
 271         git_off_t entry_start = idx->entry_start;
 272
 273         entry = git__calloc(1, sizeof(*entry));
 274         GITERR_CHECK_ALLOC(entry);
 275
 276         pentry = git__calloc(1, sizeof(struct git_pack_entry));
 277         GITERR_CHECK_ALLOC(pentry);
 278
 279         git_hash_final(&oid, &idx->hash_ctx);
 280         entry_size = idx->off - entry_start;
 281         if (entry_start > UINT31_MAX) {
 282                 entry->offset = UINT32_MAX;
 283                 entry->offset_long = entry_start;
 284         } else {
 285                 entry->offset = (uint32_t)entry_start;
 286         }
 287
 288         git_oid_cpy(&pentry->sha1, &oid);
 289         pentry->offset = entry_start;
 290
 291         k = kh_put(oid, idx->pack->idx_cache, &pentry->sha1, &error);
 292         if (error == -1) {
 293                 git__free(pentry);
 294                 giterr_set_oom();
 295                 goto on_error;
 296         }
 297
 298         if (error == 0) {
 299                 giterr_set(GITERR_INDEXER, "duplicate object %s found in pack", git_oid_tostr_s(&pentry->sha1));
 300                 git__free(pentry);
 301                 goto on_error;
 302         }
 303
 304
 305         kh_value(idx->pack->idx_cache, k) = pentry;
 306
 307         git_oid_cpy(&entry->oid, &oid);
 308
 309         if (crc_object(&entry->crc, &idx->pack->mwf, entry_start, entry_size) < 0)
 310                 goto on_error;
 311
 312         /* Add the object to the list */
 313         if (git_vector_insert(&idx->objects, entry) < 0)
 314                 goto on_error;
 315
 316         for (i = oid.id[0]; i < 256; ++i) {
 317                 idx->fanout[i]++;
 318         }
 319
 320         return 0;
 321
 322 on_error:
 323         git__free(entry);
 324
 325         return -1;
 326 }
 327
 328 GIT_INLINE(bool) has_entry(git_indexer *idx, git_oid *id)
 329 {
 330         khiter_t k;
 331         k = kh_get(oid, idx->pack->idx_cache, id);
 332         return (k != kh_end(idx->pack->idx_cache));
 333 }
 334
 335 static int save_entry(git_indexer *idx, struct entry *entry, struct git_pack_entry *pentry, git_off_t entry_start)
 336 {
 337         int i, error;
 338         khiter_t k;
 339
 340         if (entry_start > UINT31_MAX) {
 341                 entry->offset = UINT32_MAX;
 342                 entry->offset_long = entry_start;
 343         } else {
 344                 entry->offset = (uint32_t)entry_start;
 345         }
 346
 347         pentry->offset = entry_start;
 348         k = kh_put(oid, idx->pack->idx_cache, &pentry->sha1, &error);
 349
 350         if (error <= 0) {
 351                 giterr_set(GITERR_INDEXER, "cannot insert object into pack");
 352                 return -1;
 353         }
 354
 355         kh_value(idx->pack->idx_cache, k) = pentry;
 356
 357         /* Add the object to the list */
 358         if (git_vector_insert(&idx->objects, entry) < 0)
 359                 return -1;
 360
 361         for (i = entry->oid.id[0]; i < 256; ++i) {
 362                 idx->fanout[i]++;
 363         }
 364
 365         return 0;
 366 }
 367
 368 static int hash_and_save(git_indexer *idx, git_rawobj *obj, git_off_t entry_start)
 369 {
 370         git_oid oid;
 371         size_t entry_size;
 372         struct entry *entry;
 373         struct git_pack_entry *pentry = NULL;
 374
 375         entry = git__calloc(1, sizeof(*entry));
 376         GITERR_CHECK_ALLOC(entry);
 377
 378         if (git_odb__hashobj(&oid, obj) < 0) {
 379                 giterr_set(GITERR_INDEXER, "Failed to hash object");
 380                 goto on_error;
 381         }
 382
 383         pentry = git__calloc(1, sizeof(struct git_pack_entry));
 384         GITERR_CHECK_ALLOC(pentry);
 385
 386         git_oid_cpy(&pentry->sha1, &oid);
 387         git_oid_cpy(&entry->oid, &oid);
 388         entry->crc = crc32(0L, Z_NULL, 0);
 389
 390         entry_size = (size_t)(idx->off - entry_start);
 391         if (crc_object(&entry->crc, &idx->pack->mwf, entry_start, entry_size) < 0)
 392                 goto on_error;
 393
 394         return save_entry(idx, entry, pentry, entry_start);
 395
 396 on_error:
 397         git__free(pentry);
 398         git__free(entry);
 399         git__free(obj->data);
 400         return -1;
 401 }
 402
 403 static int do_progress_callback(git_indexer *idx, git_transfer_progress *stats)
 404 {
 405         if (idx->progress_cb)
 406                 return giterr_set_after_callback_function(
 407                         idx->progress_cb(stats, idx->progress_payload),
 408                         "indexer progress");
 409         return 0;
 410 }
 411
 412 /* Hash everything but the last 20B of input */
 413 static void hash_partially(git_indexer *idx, const uint8_t *data, size_t size)
 414 {
 415         size_t to_expell, to_keep;
 416
 417         if (size == 0)
 418                 return;
 419
 420         /* Easy case, dump the buffer and the data minus the last 20 bytes */
 421         if (size >= GIT_OID_RAWSZ) {
 422                 git_hash_update(&idx->trailer, idx->inbuf, idx->inbuf_len);
 423                 git_hash_update(&idx->trailer, data, size - GIT_OID_RAWSZ);
 424
 425                 data += size - GIT_OID_RAWSZ;
 426                 memcpy(idx->inbuf, data, GIT_OID_RAWSZ);
 427                 idx->inbuf_len = GIT_OID_RAWSZ;
 428                 return;
 429         }
 430
 431         /* We can just append */
 432         if (idx->inbuf_len + size <= GIT_OID_RAWSZ) {
 433                 memcpy(idx->inbuf + idx->inbuf_len, data, size);
 434                 idx->inbuf_len += size;
 435                 return;
 436         }
 437
 438         /* We need to partially drain the buffer and then append */
 439         to_keep   = GIT_OID_RAWSZ - size;
 440         to_expell = idx->inbuf_len - to_keep;
 441
 442         git_hash_update(&idx->trailer, idx->inbuf, to_expell);
 443
 444         memmove(idx->inbuf, idx->inbuf + to_expell, to_keep);
 445         memcpy(idx->inbuf + to_keep, data, size);
 446         idx->inbuf_len += size - to_expell;
 447 }
 448
 449 static int write_at(git_indexer *idx, const void *data, git_off_t offset, size_t size)
 450 {
 451         git_file fd = idx->pack->mwf.fd;
 452         size_t mmap_alignment;
 453         size_t page_offset;
 454         git_off_t page_start;
 455         unsigned char *map_data;
 456         git_map map;
 457         int error;
 458
 459         assert(data && size);
 460
 461         if ((error = git__mmap_alignment(&mmap_alignment)) < 0)
 462                 return error;
 463
 464         /* the offset needs to be at the mmap boundary for the platform */
 465         page_offset = offset % mmap_alignment;
 466         page_start = offset - page_offset;
 467
 468         if ((error = p_mmap(&map, page_offset + size, GIT_PROT_WRITE, GIT_MAP_SHARED, fd, page_start)) < 0)
 469                 return error;
 470
 471         map_data = (unsigned char *)map.data;
 472         memcpy(map_data + page_offset, data, size);
 473         p_munmap(&map);
 474
 475         return 0;
 476 }
 477
 478 static int append_to_pack(git_indexer *idx, const void *data, size_t size)
 479 {
 480         git_off_t new_size;
 481         size_t mmap_alignment;
 482         size_t page_offset;
 483         git_off_t page_start;
 484         git_off_t current_size = idx->pack->mwf.size;
 485         int fd = idx->pack->mwf.fd;
 486         int error;
 487
 488         if (!size)
 489                 return 0;
 490
 491         if ((error = git__mmap_alignment(&mmap_alignment)) < 0)
 492                 return error;
 493
 494         /* Write a single byte to force the file system to allocate space now or
 495          * report an error, since we can't report errors when writing using mmap.
 496          * Round the size up to the nearest page so that we only need to perform file
 497          * I/O when we add a page, instead of whenever we write even a single byte. */
 498         new_size = current_size + size;
 499         page_offset = new_size % mmap_alignment;
 500         page_start = new_size - page_offset;
 501
 502         if (p_lseek(fd, page_start + mmap_alignment - 1, SEEK_SET) < 0 ||
 503             p_write(idx->pack->mwf.fd, data, 1) < 0) {
 504                 giterr_set(GITERR_OS, "cannot extend packfile '%s'", idx->pack->pack_name);
 505                 return -1;
 506         }
 507
 508         return write_at(idx, data, idx->pack->mwf.size, size);
 509 }
 510
 511 int git_indexer_append(git_indexer *idx, const void *data, size_t size, git_transfer_progress *stats)
 512 {
 513         int error = -1;
 514         size_t processed;
 515         struct git_pack_header *hdr = &idx->hdr;
 516         git_mwindow_file *mwf = &idx->pack->mwf;
 517
 518         assert(idx && data && stats);
 519
 520         processed = stats->indexed_objects;
 521
 522         if ((error = append_to_pack(idx, data, size)) < 0)
 523                 return error;
 524
 525         hash_partially(idx, data, (int)size);
 526
 527         /* Make sure we set the new size of the pack */
 528         idx->pack->mwf.size += size;
 529
 530         if (!idx->parsed_header) {
 531                 unsigned int total_objects;
 532
 533                 if ((unsigned)idx->pack->mwf.size < sizeof(struct git_pack_header))
 534                         return 0;
 535
 536                 if ((error = parse_header(&idx->hdr, idx->pack)) < 0)
 537                         return error;
 538
 539                 idx->parsed_header = 1;
 540                 idx->nr_objects = ntohl(hdr->hdr_entries);
 541                 idx->off = sizeof(struct git_pack_header);
 542
 543                 /* for now, limit to 2^32 objects */
 544                 assert(idx->nr_objects == (size_t)((unsigned int)idx->nr_objects));
 545                 if (idx->nr_objects == (size_t)((unsigned int)idx->nr_objects))
 546                         total_objects = (unsigned int)idx->nr_objects;
 547                 else
 548                         total_objects = UINT_MAX;
 549
 550                 idx->pack->idx_cache = git_oidmap_alloc();
 551                 GITERR_CHECK_ALLOC(idx->pack->idx_cache);
 552
 553                 idx->pack->has_cache = 1;
 554                 if (git_vector_init(&idx->objects, total_objects, objects_cmp) < 0)
 555                         return -1;
 556
 557                 if (git_vector_init(&idx->deltas, total_objects / 2, NULL) < 0)
 558                         return -1;
 559
 560                 stats->received_objects = 0;
 561                 stats->local_objects = 0;
 562                 stats->total_deltas = 0;
 563                 stats->indexed_deltas = 0;
 564                 processed = stats->indexed_objects = 0;
 565                 stats->total_objects = total_objects;
 566
 567                 if ((error = do_progress_callback(idx, stats)) != 0)
 568                         return error;
 569         }
 570
 571         /* Now that we have data in the pack, let's try to parse it */
 572
 573         /* As the file grows any windows we try to use will be out of date */
 574         git_mwindow_free_all(mwf);
 575
 576         while (processed < idx->nr_objects) {
 577                 git_packfile_stream *stream = &idx->stream;
 578                 git_off_t entry_start = idx->off;
 579                 size_t entry_size;
 580                 git_otype type;
 581                 git_mwindow *w = NULL;
 582
 583                 if (idx->pack->mwf.size <= idx->off + 20)
 584                         return 0;
 585
 586                 if (!idx->have_stream) {
 587                         error = git_packfile_unpack_header(&entry_size, &type, mwf, &w, &idx->off);
 588                         if (error == GIT_EBUFS) {
 589                                 idx->off = entry_start;
 590                                 return 0;
 591                         }
 592                         if (error < 0)
 593                                 goto on_error;
 594
 595                         git_mwindow_close(&w);
 596                         idx->entry_start = entry_start;
 597                         git_hash_init(&idx->hash_ctx);
 598
 599                         if (type == GIT_OBJ_REF_DELTA || type == GIT_OBJ_OFS_DELTA) {
 600                                 error = advance_delta_offset(idx, type);
 601                                 if (error == GIT_EBUFS) {
 602                                         idx->off = entry_start;
 603                                         return 0;
 604                                 }
 605                                 if (error < 0)
 606                                         goto on_error;
 607
 608                                 idx->have_delta = 1;
 609                         } else {
 610                                 idx->have_delta = 0;
 611                                 hash_header(&idx->hash_ctx, entry_size, type);
 612                         }
 613
 614                         idx->have_stream = 1;
 615
 616                         error = git_packfile_stream_open(stream, idx->pack, idx->off);
 617                         if (error < 0)
 618                                 goto on_error;
 619                 }
 620
 621                 if (idx->have_delta) {
 622                         error = read_object_stream(idx, stream);
 623                 } else {
 624                         error = hash_object_stream(idx, stream);
 625                 }
 626
 627                 idx->off = stream->curpos;
 628                 if (error == GIT_EBUFS)
 629                         return 0;
 630
 631                 /* We want to free the stream reasorces no matter what here */
 632                 idx->have_stream = 0;
 633                 git_packfile_stream_free(stream);
 634
 635                 if (error < 0)
 636                         goto on_error;
 637
 638                 if (idx->have_delta) {
 639                         error = store_delta(idx);
 640                 } else {
 641                         error = store_object(idx);
 642                 }
 643
 644                 if (error < 0)
 645                         goto on_error;
 646
 647                 if (!idx->have_delta) {
 648                         stats->indexed_objects = (unsigned int)++processed;
 649                 }
 650                 stats->received_objects++;
 651
 652                 if ((error = do_progress_callback(idx, stats)) != 0)
 653                         goto on_error;
 654         }
 655
 656         return 0;
 657
 658 on_error:
 659         git_mwindow_free_all(mwf);
 660         return error;
 661 }
 662
 663 static int index_path(git_buf *path, git_indexer *idx, const char *suffix)
 664 {
 665         const char prefix[] = "pack-";
 666         size_t slash = (size_t)path->size;
 667
 668         /* search backwards for '/' */
 669         while (slash > 0 && path->ptr[slash - 1] != '/')
 670                 slash--;
 671
 672         if (git_buf_grow(path, slash + 1 + strlen(prefix) +
 673                                          GIT_OID_HEXSZ + strlen(suffix) + 1) < 0)
 674                 return -1;
 675
 676         git_buf_truncate(path, slash);
 677         git_buf_puts(path, prefix);
 678         git_oid_fmt(path->ptr + git_buf_len(path), &idx->hash);
 679         path->size += GIT_OID_HEXSZ;
 680         git_buf_puts(path, suffix);
 681
 682         return git_buf_oom(path) ? -1 : 0;
 683 }
 684
 685 /**
 686  * Rewind the packfile by the trailer, as we might need to fix the
 687  * packfile by injecting objects at the tail and must overwrite it.
 688  */
 689 static void seek_back_trailer(git_indexer *idx)
 690 {
 691         idx->pack->mwf.size -= GIT_OID_RAWSZ;
 692         git_mwindow_free_all(&idx->pack->mwf);
 693 }
 694
 695 static int inject_object(git_indexer *idx, git_oid *id)
 696 {
 697         git_odb_object *obj;
 698         struct entry *entry;
 699         struct git_pack_entry *pentry = NULL;
 700         git_oid foo = {{0}};
 701         unsigned char hdr[64];
 702         git_buf buf = GIT_BUF_INIT;
 703         git_off_t entry_start;
 704         const void *data;
 705         size_t len, hdr_len;
 706         int error;
 707
 708         seek_back_trailer(idx);
 709         entry_start = idx->pack->mwf.size;
 710
 711         if (git_odb_read(&obj, idx->odb, id) < 0) {
 712                 giterr_set(GITERR_INDEXER, "missing delta bases");
 713                 return -1;
 714         }
 715
 716         data = git_odb_object_data(obj);
 717         len = git_odb_object_size(obj);
 718
 719         entry = git__calloc(1, sizeof(*entry));
 720         GITERR_CHECK_ALLOC(entry);
 721
 722         entry->crc = crc32(0L, Z_NULL, 0);
 723
 724         /* Write out the object header */
 725         hdr_len = git_packfile__object_header(hdr, len, git_odb_object_type(obj));
 726         if ((error = append_to_pack(idx, hdr, hdr_len)) < 0)
 727                 goto cleanup;
 728
 729         idx->pack->mwf.size += hdr_len;
 730         entry->crc = crc32(entry->crc, hdr, (uInt)hdr_len);
 731
 732         if ((error = git_zstream_deflatebuf(&buf, data, len)) < 0)
 733                 goto cleanup;
 734
 735         /* And then the compressed object */
 736         if ((error = append_to_pack(idx, buf.ptr, buf.size)) < 0)
 737                 goto cleanup;
 738
 739         idx->pack->mwf.size += buf.size;
 740         entry->crc = htonl(crc32(entry->crc, (unsigned char *)buf.ptr, (uInt)buf.size));
 741         git_buf_free(&buf);
 742
 743         /* Write a fake trailer so the pack functions play ball */
 744
 745         if ((error = append_to_pack(idx, &foo, GIT_OID_RAWSZ)) < 0)
 746                 goto cleanup;
 747
 748         idx->pack->mwf.size += GIT_OID_RAWSZ;
 749
 750         pentry = git__calloc(1, sizeof(struct git_pack_entry));
 751         GITERR_CHECK_ALLOC(pentry);
 752
 753         git_oid_cpy(&pentry->sha1, id);
 754         git_oid_cpy(&entry->oid, id);
 755         idx->off = entry_start + hdr_len + len;
 756
 757         error = save_entry(idx, entry, pentry, entry_start);
 758
 759 cleanup:
 760         if (error) {
 761                 git__free(entry);
 762                 git__free(pentry);
 763         }
 764
 765         git_odb_object_free(obj);
 766         return error;
 767 }
 768
 769 static int fix_thin_pack(git_indexer *idx, git_transfer_progress *stats)
 770 {
 771         int error, found_ref_delta = 0;
 772         unsigned int i;
 773         struct delta_info *delta;
 774         size_t size;
 775         git_otype type;
 776         git_mwindow *w = NULL;
 777         git_off_t curpos = 0;
 778         unsigned char *base_info;
 779         unsigned int left = 0;
 780         git_oid base;
 781
 782         assert(git_vector_length(&idx->deltas) > 0);
 783
 784         if (idx->odb == NULL) {
 785                 giterr_set(GITERR_INDEXER, "cannot fix a thin pack without an ODB");
 786                 return -1;
 787         }
 788
 789         /* Loop until we find the first REF delta */
 790         git_vector_foreach(&idx->deltas, i, delta) {
 791                 if (!delta)
 792                         continue;
 793
 794                 curpos = delta->delta_off;
 795                 error = git_packfile_unpack_header(&size, &type, &idx->pack->mwf, &w, &curpos);
 796                 if (error < 0)
 797                         return error;
 798
 799                 if (type == GIT_OBJ_REF_DELTA) {
 800                         found_ref_delta = 1;
 801                         break;
 802                 }
 803         }
 804
 805         if (!found_ref_delta) {
 806                 giterr_set(GITERR_INDEXER, "no REF_DELTA found, cannot inject object");
 807                 return -1;
 808         }
 809
 810         /* curpos now points to the base information, which is an OID */
 811         base_info = git_mwindow_open(&idx->pack->mwf, &w, curpos, GIT_OID_RAWSZ, &left);
 812         if (base_info == NULL) {
 813                 giterr_set(GITERR_INDEXER, "failed to map delta information");
 814                 return -1;
 815         }
 816
 817         git_oid_fromraw(&base, base_info);
 818         git_mwindow_close(&w);
 819
 820         if (has_entry(idx, &base))
 821                 return 0;
 822
 823         if (inject_object(idx, &base) < 0)
 824                 return -1;
 825
 826         stats->local_objects++;
 827
 828         return 0;
 829 }
 830
 831 static int resolve_deltas(git_indexer *idx, git_transfer_progress *stats)
 832 {
 833         unsigned int i;
 834         struct delta_info *delta;
 835         int progressed = 0, non_null = 0, progress_cb_result;
 836
 837         while (idx->deltas.length > 0) {
 838                 progressed = 0;
 839                 non_null = 0;
 840                 git_vector_foreach(&idx->deltas, i, delta) {
 841                         git_rawobj obj = {NULL};
 842
 843                         if (!delta)
 844                                 continue;
 845
 846                         non_null = 1;
 847                         idx->off = delta->delta_off;
 848                         if (git_packfile_unpack(&obj, idx->pack, &idx->off) < 0)
 849                                 continue;
 850
 851                         if (hash_and_save(idx, &obj, delta->delta_off) < 0)
 852                                 continue;
 853
 854                         git__free(obj.data);
 855                         stats->indexed_objects++;
 856                         stats->indexed_deltas++;
 857                         progressed = 1;
 858                         if ((progress_cb_result = do_progress_callback(idx, stats)) < 0)
 859                                 return progress_cb_result;
 860
 861                         /* remove from the list */
 862                         git_vector_set(NULL, &idx->deltas, i, NULL);
 863                         git__free(delta);
 864                 }
 865
 866                 /* if none were actually set, we're done */
 867                 if (!non_null)
 868                         break;
 869
 870                 if (!progressed && (fix_thin_pack(idx, stats) < 0)) {
 871                         return -1;
 872                 }
 873         }
 874
 875         return 0;
 876 }
 877
 878 static int update_header_and_rehash(git_indexer *idx, git_transfer_progress *stats)
 879 {
 880         void *ptr;
 881         size_t chunk = 1024*1024;
 882         git_off_t hashed = 0;
 883         git_mwindow *w = NULL;
 884         git_mwindow_file *mwf;
 885         unsigned int left;
 886
 887         mwf = &idx->pack->mwf;
 888
 889         git_hash_init(&idx->trailer);
 890
 891
 892         /* Update the header to include the numer of local objects we injected */
 893         idx->hdr.hdr_entries = htonl(stats->total_objects + stats->local_objects);
 894         if (write_at(idx, &idx->hdr, 0, sizeof(struct git_pack_header)) < 0)
 895                 return -1;
 896
 897         /*
 898          * We now use the same technique as before to determine the
 899          * hash. We keep reading up to the end and let
 900          * hash_partially() keep the existing trailer out of the
 901          * calculation.
 902          */
 903         git_mwindow_free_all(mwf);
 904         idx->inbuf_len = 0;
 905         while (hashed < mwf->size) {
 906                 ptr = git_mwindow_open(mwf, &w, hashed, chunk, &left);
 907                 if (ptr == NULL)
 908                         return -1;
 909
 910                 hash_partially(idx, ptr, left);
 911                 hashed += left;
 912
 913                 git_mwindow_close(&w);
 914         }
 915
 916         return 0;
 917 }
 918
 919 int git_indexer_commit(git_indexer *idx, git_transfer_progress *stats)
 920 {
 921         git_mwindow *w = NULL;
 922         unsigned int i, long_offsets = 0, left;
 923         int error;
 924         struct git_pack_idx_header hdr;
 925         git_buf filename = GIT_BUF_INIT;
 926         struct entry *entry;
 927         git_oid trailer_hash, file_hash;
 928         git_hash_ctx ctx;
 929         git_filebuf index_file = {0};
 930         void *packfile_trailer;
 931
 932         if (!idx->parsed_header) {
 933                 giterr_set(GITERR_INDEXER, "incomplete pack header");
 934                 return -1;
 935         }
 936
 937         if (git_hash_ctx_init(&ctx) < 0)
 938                 return -1;
 939
 940         /* Test for this before resolve_deltas(), as it plays with idx->off */
 941         if (idx->off + 20 < idx->pack->mwf.size) {
 942                 giterr_set(GITERR_INDEXER, "unexpected data at the end of the pack");
 943                 return -1;
 944         }
 945
 946         packfile_trailer = git_mwindow_open(&idx->pack->mwf, &w, idx->pack->mwf.size - GIT_OID_RAWSZ, GIT_OID_RAWSZ, &left);
 947         if (packfile_trailer == NULL) {
 948                 git_mwindow_close(&w);
 949                 goto on_error;
 950         }
 951
 952         /* Compare the packfile trailer as it was sent to us and what we calculated */
 953         git_oid_fromraw(&file_hash, packfile_trailer);
 954         git_mwindow_close(&w);
 955
 956         git_hash_final(&trailer_hash, &idx->trailer);
 957         if (git_oid_cmp(&file_hash, &trailer_hash)) {
 958                 giterr_set(GITERR_INDEXER, "packfile trailer mismatch");
 959                 return -1;
 960         }
 961
 962         /* Freeze the number of deltas */
 963         stats->total_deltas = stats->total_objects - stats->indexed_objects;
 964
 965         if ((error = resolve_deltas(idx, stats)) < 0)
 966                 return error;
 967
 968         if (stats->indexed_objects != stats->total_objects) {
 969                 giterr_set(GITERR_INDEXER, "early EOF");
 970                 return -1;
 971         }
 972
 973         if (stats->local_objects > 0) {
 974                 if (update_header_and_rehash(idx, stats) < 0)
 975                         return -1;
 976
 977                 git_hash_final(&trailer_hash, &idx->trailer);
 978                 write_at(idx, &trailer_hash, idx->pack->mwf.size - GIT_OID_RAWSZ, GIT_OID_RAWSZ);
 979         }
 980
 981         git_vector_sort(&idx->objects);
 982
 983         git_buf_sets(&filename, idx->pack->pack_name);
 984         git_buf_shorten(&filename, strlen("pack"));
 985         git_buf_puts(&filename, "idx");
 986         if (git_buf_oom(&filename))
 987                 return -1;
 988
 989         if (git_filebuf_open(&index_file, filename.ptr,
 990                 GIT_FILEBUF_HASH_CONTENTS, idx->mode) < 0)
 991                 goto on_error;
 992
 993         /* Write out the header */
 994         hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
 995         hdr.idx_version = htonl(2);
 996         git_filebuf_write(&index_file, &hdr, sizeof(hdr));
 997
 998         /* Write out the fanout table */
 999         for (i = 0; i < 256; ++i) {
1000                 uint32_t n = htonl(idx->fanout[i]);
1001                 git_filebuf_write(&index_file, &n, sizeof(n));
1002         }
1003
1004         /* Write out the object names (SHA-1 hashes) */
1005         git_vector_foreach(&idx->objects, i, entry) {
1006                 git_filebuf_write(&index_file, &entry->oid, sizeof(git_oid));
1007                 git_hash_update(&ctx, &entry->oid, GIT_OID_RAWSZ);
1008         }
1009         git_hash_final(&idx->hash, &ctx);
1010
1011         /* Write out the CRC32 values */
1012         git_vector_foreach(&idx->objects, i, entry) {
1013                 git_filebuf_write(&index_file, &entry->crc, sizeof(uint32_t));
1014         }
1015
1016         /* Write out the offsets */
1017         git_vector_foreach(&idx->objects, i, entry) {
1018                 uint32_t n;
1019
1020                 if (entry->offset == UINT32_MAX)
1021                         n = htonl(0x80000000 | long_offsets++);
1022                 else
1023                         n = htonl(entry->offset);
1024
1025                 git_filebuf_write(&index_file, &n, sizeof(uint32_t));
1026         }
1027
1028         /* Write out the long offsets */
1029         git_vector_foreach(&idx->objects, i, entry) {
1030                 uint32_t split[2];
1031
1032                 if (entry->offset != UINT32_MAX)
1033                         continue;
1034
1035                 split[0] = htonl(entry->offset_long >> 32);
1036                 split[1] = htonl(entry->offset_long & 0xffffffff);
1037
1038                 git_filebuf_write(&index_file, &split, sizeof(uint32_t) * 2);
1039         }
1040
1041         /* Write out the packfile trailer to the index */
1042         if (git_filebuf_write(&index_file, &trailer_hash, GIT_OID_RAWSZ) < 0)
1043                 goto on_error;
1044
1045         /* Write out the hash of the idx */
1046         if (git_filebuf_hash(&trailer_hash, &index_file) < 0)
1047                 goto on_error;
1048
1049         git_filebuf_write(&index_file, &trailer_hash, sizeof(git_oid));
1050
1051         /* Figure out what the final name should be */
1052         if (index_path(&filename, idx, ".idx") < 0)
1053                 goto on_error;
1054
1055         /* Commit file */
1056         if (git_filebuf_commit_at(&index_file, filename.ptr) < 0)
1057                 goto on_error;
1058
1059         git_mwindow_free_all(&idx->pack->mwf);
1060
1061         /* Truncate file to undo rounding up to next page_size in append_to_pack */
1062         if (p_ftruncate(idx->pack->mwf.fd, idx->pack->mwf.size) < 0) {
1063                 giterr_set(GITERR_OS, "failed to truncate pack file '%s'", idx->pack->pack_name);
1064                 return -1;
1065         }
1066
1067         /* We need to close the descriptor here so Windows doesn't choke on commit_at */
1068         if (p_close(idx->pack->mwf.fd) < 0) {
1069                 giterr_set(GITERR_OS, "failed to close packfile");
1070                 goto on_error;
1071         }
1072
1073         idx->pack->mwf.fd = -1;
1074
1075         if (index_path(&filename, idx, ".pack") < 0)
1076                 goto on_error;
1077
1078         /* And don't forget to rename the packfile to its new place. */
1079         p_rename(idx->pack->pack_name, git_buf_cstr(&filename));
1080
1081         git_buf_free(&filename);
1082         git_hash_ctx_cleanup(&ctx);
1083         return 0;
1084
1085 on_error:
1086         git_mwindow_free_all(&idx->pack->mwf);
1087         git_filebuf_cleanup(&index_file);
1088         git_buf_free(&filename);
1089         git_hash_ctx_cleanup(&ctx);
1090         return -1;
1091 }
1092
1093 void git_indexer_free(git_indexer *idx)
1094 {
1095         if (idx == NULL)
1096                 return;
1097
1098         git_vector_free_deep(&idx->objects);
1099
1100         if (idx->pack && idx->pack->idx_cache) {
1101                 struct git_pack_entry *pentry;
1102                 kh_foreach_value(
1103                         idx->pack->idx_cache, pentry, { git__free(pentry); });
1104
1105                 git_oidmap_free(idx->pack->idx_cache);
1106         }
1107
1108         git_vector_free_deep(&idx->deltas);
1109
1110         if (!git_mutex_lock(&git__mwindow_mutex)) {
1111                 git_packfile_free(idx->pack);
1112                 git_mutex_unlock(&git__mwindow_mutex);
1113         }
1114
1115         git_hash_ctx_cleanup(&idx->trailer);
1116         git_hash_ctx_cleanup(&idx->hash_ctx);
1117         git__free(idx);
1118 }