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 static int save_entry(git_indexer *idx, struct entry *entry, struct git_pack_entry *pentry, git_off_t entry_start)
 329 {
 330         int i, error;
 331         khiter_t k;
 332
 333         if (entry_start > UINT31_MAX) {
 334                 entry->offset = UINT32_MAX;
 335                 entry->offset_long = entry_start;
 336         } else {
 337                 entry->offset = (uint32_t)entry_start;
 338         }
 339
 340         pentry->offset = entry_start;
 341         k = kh_put(oid, idx->pack->idx_cache, &pentry->sha1, &error);
 342         if (!error)
 343                 return -1;
 344
 345         kh_value(idx->pack->idx_cache, k) = pentry;
 346
 347         /* Add the object to the list */
 348         if (git_vector_insert(&idx->objects, entry) < 0)
 349                 return -1;
 350
 351         for (i = entry->oid.id[0]; i < 256; ++i) {
 352                 idx->fanout[i]++;
 353         }
 354
 355         return 0;
 356 }
 357
 358 static int hash_and_save(git_indexer *idx, git_rawobj *obj, git_off_t entry_start)
 359 {
 360         git_oid oid;
 361         size_t entry_size;
 362         struct entry *entry;
 363         struct git_pack_entry *pentry = NULL;
 364
 365         entry = git__calloc(1, sizeof(*entry));
 366         GITERR_CHECK_ALLOC(entry);
 367
 368         if (git_odb__hashobj(&oid, obj) < 0) {
 369                 giterr_set(GITERR_INDEXER, "Failed to hash object");
 370                 goto on_error;
 371         }
 372
 373         pentry = git__calloc(1, sizeof(struct git_pack_entry));
 374         GITERR_CHECK_ALLOC(pentry);
 375
 376         git_oid_cpy(&pentry->sha1, &oid);
 377         git_oid_cpy(&entry->oid, &oid);
 378         entry->crc = crc32(0L, Z_NULL, 0);
 379
 380         entry_size = (size_t)(idx->off - entry_start);
 381         if (crc_object(&entry->crc, &idx->pack->mwf, entry_start, entry_size) < 0)
 382                 goto on_error;
 383
 384         return save_entry(idx, entry, pentry, entry_start);
 385
 386 on_error:
 387         git__free(pentry);
 388         git__free(entry);
 389         git__free(obj->data);
 390         return -1;
 391 }
 392
 393 static int do_progress_callback(git_indexer *idx, git_transfer_progress *stats)
 394 {
 395         if (idx->progress_cb)
 396                 return giterr_set_after_callback_function(
 397                         idx->progress_cb(stats, idx->progress_payload),
 398                         "indexer progress");
 399         return 0;
 400 }
 401
 402 /* Hash everything but the last 20B of input */
 403 static void hash_partially(git_indexer *idx, const uint8_t *data, size_t size)
 404 {
 405         size_t to_expell, to_keep;
 406
 407         if (size == 0)
 408                 return;
 409
 410         /* Easy case, dump the buffer and the data minus the last 20 bytes */
 411         if (size >= GIT_OID_RAWSZ) {
 412                 git_hash_update(&idx->trailer, idx->inbuf, idx->inbuf_len);
 413                 git_hash_update(&idx->trailer, data, size - GIT_OID_RAWSZ);
 414
 415                 data += size - GIT_OID_RAWSZ;
 416                 memcpy(idx->inbuf, data, GIT_OID_RAWSZ);
 417                 idx->inbuf_len = GIT_OID_RAWSZ;
 418                 return;
 419         }
 420
 421         /* We can just append */
 422         if (idx->inbuf_len + size <= GIT_OID_RAWSZ) {
 423                 memcpy(idx->inbuf + idx->inbuf_len, data, size);
 424                 idx->inbuf_len += size;
 425                 return;
 426         }
 427
 428         /* We need to partially drain the buffer and then append */
 429         to_keep   = GIT_OID_RAWSZ - size;
 430         to_expell = idx->inbuf_len - to_keep;
 431
 432         git_hash_update(&idx->trailer, idx->inbuf, to_expell);
 433
 434         memmove(idx->inbuf, idx->inbuf + to_expell, to_keep);
 435         memcpy(idx->inbuf + to_keep, data, size);
 436         idx->inbuf_len += size - to_expell;
 437 }
 438
 439 static int write_at(git_indexer *idx, const void *data, git_off_t offset, size_t size)
 440 {
 441         git_file fd = idx->pack->mwf.fd;
 442         size_t page_size;
 443         size_t page_offset;
 444         git_off_t page_start;
 445         unsigned char *map_data;
 446         git_map map;
 447         int error;
 448
 449         assert(data && size);
 450
 451         if ((error = git__page_size(&page_size)) < 0)
 452                 return error;
 453
 454         /* the offset needs to be at the beginning of the a page boundary */
 455         page_offset = offset % page_size;
 456         page_start = offset - page_offset;
 457
 458         if ((error = p_mmap(&map, page_offset + size, GIT_PROT_WRITE, GIT_MAP_SHARED, fd, page_start)) < 0)
 459                 return error;
 460
 461         map_data = (unsigned char *)map.data;
 462         memcpy(map_data + page_offset, data, size);
 463         p_munmap(&map);
 464
 465         return 0;
 466 }
 467
 468 static int append_to_pack(git_indexer *idx, const void *data, size_t size)
 469 {
 470         git_off_t current_size = idx->pack->mwf.size;
 471
 472         if (!size)
 473                 return 0;
 474
 475         /* add the extra space we need at the end */
 476         if (p_ftruncate(idx->pack->mwf.fd, current_size + size) < 0) {
 477                 giterr_set(GITERR_OS, "Failed to increase size of pack file '%s'", idx->pack->pack_name);
 478                 return -1;
 479         }
 480
 481         return write_at(idx, data, idx->pack->mwf.size, size);
 482 }
 483
 484 int git_indexer_append(git_indexer *idx, const void *data, size_t size, git_transfer_progress *stats)
 485 {
 486         int error = -1;
 487         size_t processed;
 488         struct git_pack_header *hdr = &idx->hdr;
 489         git_mwindow_file *mwf = &idx->pack->mwf;
 490
 491         assert(idx && data && stats);
 492
 493         processed = stats->indexed_objects;
 494
 495         if ((error = append_to_pack(idx, data, size)) < 0)
 496                 return error;
 497
 498         hash_partially(idx, data, (int)size);
 499
 500         /* Make sure we set the new size of the pack */
 501         idx->pack->mwf.size += size;
 502
 503         if (!idx->parsed_header) {
 504                 unsigned int total_objects;
 505
 506                 if ((unsigned)idx->pack->mwf.size < sizeof(struct git_pack_header))
 507                         return 0;
 508
 509                 if ((error = parse_header(&idx->hdr, idx->pack)) < 0)
 510                         return error;
 511
 512                 idx->parsed_header = 1;
 513                 idx->nr_objects = ntohl(hdr->hdr_entries);
 514                 idx->off = sizeof(struct git_pack_header);
 515
 516                 /* for now, limit to 2^32 objects */
 517                 assert(idx->nr_objects == (size_t)((unsigned int)idx->nr_objects));
 518                 if (idx->nr_objects == (size_t)((unsigned int)idx->nr_objects))
 519                         total_objects = (unsigned int)idx->nr_objects;
 520                 else
 521                         total_objects = UINT_MAX;
 522
 523                 idx->pack->idx_cache = git_oidmap_alloc();
 524                 GITERR_CHECK_ALLOC(idx->pack->idx_cache);
 525
 526                 idx->pack->has_cache = 1;
 527                 if (git_vector_init(&idx->objects, total_objects, objects_cmp) < 0)
 528                         return -1;
 529
 530                 if (git_vector_init(&idx->deltas, total_objects / 2, NULL) < 0)
 531                         return -1;
 532
 533                 stats->received_objects = 0;
 534                 stats->local_objects = 0;
 535                 stats->total_deltas = 0;
 536                 stats->indexed_deltas = 0;
 537                 processed = stats->indexed_objects = 0;
 538                 stats->total_objects = total_objects;
 539
 540                 if ((error = do_progress_callback(idx, stats)) != 0)
 541                         return error;
 542         }
 543
 544         /* Now that we have data in the pack, let's try to parse it */
 545
 546         /* As the file grows any windows we try to use will be out of date */
 547         git_mwindow_free_all(mwf);
 548
 549         while (processed < idx->nr_objects) {
 550                 git_packfile_stream *stream = &idx->stream;
 551                 git_off_t entry_start = idx->off;
 552                 size_t entry_size;
 553                 git_otype type;
 554                 git_mwindow *w = NULL;
 555
 556                 if (idx->pack->mwf.size <= idx->off + 20)
 557                         return 0;
 558
 559                 if (!idx->have_stream) {
 560                         error = git_packfile_unpack_header(&entry_size, &type, mwf, &w, &idx->off);
 561                         if (error == GIT_EBUFS) {
 562                                 idx->off = entry_start;
 563                                 return 0;
 564                         }
 565                         if (error < 0)
 566                                 goto on_error;
 567
 568                         git_mwindow_close(&w);
 569                         idx->entry_start = entry_start;
 570                         git_hash_init(&idx->hash_ctx);
 571
 572                         if (type == GIT_OBJ_REF_DELTA || type == GIT_OBJ_OFS_DELTA) {
 573                                 error = advance_delta_offset(idx, type);
 574                                 if (error == GIT_EBUFS) {
 575                                         idx->off = entry_start;
 576                                         return 0;
 577                                 }
 578                                 if (error < 0)
 579                                         goto on_error;
 580
 581                                 idx->have_delta = 1;
 582                         } else {
 583                                 idx->have_delta = 0;
 584                                 hash_header(&idx->hash_ctx, entry_size, type);
 585                         }
 586
 587                         idx->have_stream = 1;
 588
 589                         error = git_packfile_stream_open(stream, idx->pack, idx->off);
 590                         if (error < 0)
 591                                 goto on_error;
 592                 }
 593
 594                 if (idx->have_delta) {
 595                         error = read_object_stream(idx, stream);
 596                 } else {
 597                         error = hash_object_stream(idx, stream);
 598                 }
 599
 600                 idx->off = stream->curpos;
 601                 if (error == GIT_EBUFS)
 602                         return 0;
 603
 604                 /* We want to free the stream reasorces no matter what here */
 605                 idx->have_stream = 0;
 606                 git_packfile_stream_free(stream);
 607
 608                 if (error < 0)
 609                         goto on_error;
 610
 611                 if (idx->have_delta) {
 612                         error = store_delta(idx);
 613                 } else {
 614                         error = store_object(idx);
 615                 }
 616
 617                 if (error < 0)
 618                         goto on_error;
 619
 620                 if (!idx->have_delta) {
 621                         stats->indexed_objects = (unsigned int)++processed;
 622                 }
 623                 stats->received_objects++;
 624
 625                 if ((error = do_progress_callback(idx, stats)) != 0)
 626                         goto on_error;
 627         }
 628
 629         return 0;
 630
 631 on_error:
 632         git_mwindow_free_all(mwf);
 633         return error;
 634 }
 635
 636 static int index_path(git_buf *path, git_indexer *idx, const char *suffix)
 637 {
 638         const char prefix[] = "pack-";
 639         size_t slash = (size_t)path->size;
 640
 641         /* search backwards for '/' */
 642         while (slash > 0 && path->ptr[slash - 1] != '/')
 643                 slash--;
 644
 645         if (git_buf_grow(path, slash + 1 + strlen(prefix) +
 646                                          GIT_OID_HEXSZ + strlen(suffix) + 1) < 0)
 647                 return -1;
 648
 649         git_buf_truncate(path, slash);
 650         git_buf_puts(path, prefix);
 651         git_oid_fmt(path->ptr + git_buf_len(path), &idx->hash);
 652         path->size += GIT_OID_HEXSZ;
 653         git_buf_puts(path, suffix);
 654
 655         return git_buf_oom(path) ? -1 : 0;
 656 }
 657
 658 /**
 659  * Rewind the packfile by the trailer, as we might need to fix the
 660  * packfile by injecting objects at the tail and must overwrite it.
 661  */
 662 static void seek_back_trailer(git_indexer *idx)
 663 {
 664         idx->pack->mwf.size -= GIT_OID_RAWSZ;
 665         git_mwindow_free_all(&idx->pack->mwf);
 666 }
 667
 668 static int inject_object(git_indexer *idx, git_oid *id)
 669 {
 670         git_odb_object *obj;
 671         struct entry *entry;
 672         struct git_pack_entry *pentry = NULL;
 673         git_oid foo = {{0}};
 674         unsigned char hdr[64];
 675         git_buf buf = GIT_BUF_INIT;
 676         git_off_t entry_start;
 677         const void *data;
 678         size_t len, hdr_len;
 679         int error;
 680
 681         seek_back_trailer(idx);
 682         entry_start = idx->pack->mwf.size;
 683
 684         if (git_odb_read(&obj, idx->odb, id) < 0) {
 685                 giterr_set(GITERR_INDEXER, "missing delta bases");
 686                 return -1;
 687         }
 688
 689         data = git_odb_object_data(obj);
 690         len = git_odb_object_size(obj);
 691
 692         entry = git__calloc(1, sizeof(*entry));
 693         GITERR_CHECK_ALLOC(entry);
 694
 695         entry->crc = crc32(0L, Z_NULL, 0);
 696
 697         /* Write out the object header */
 698         hdr_len = git_packfile__object_header(hdr, len, git_odb_object_type(obj));
 699         if ((error = append_to_pack(idx, hdr, hdr_len)) < 0)
 700                 goto cleanup;
 701
 702         idx->pack->mwf.size += hdr_len;
 703         entry->crc = crc32(entry->crc, hdr, (uInt)hdr_len);
 704
 705         if ((error = git_zstream_deflatebuf(&buf, data, len)) < 0)
 706                 goto cleanup;
 707
 708         /* And then the compressed object */
 709         if ((error = append_to_pack(idx, buf.ptr, buf.size)) < 0)
 710                 goto cleanup;
 711
 712         idx->pack->mwf.size += buf.size;
 713         entry->crc = htonl(crc32(entry->crc, (unsigned char *)buf.ptr, (uInt)buf.size));
 714         git_buf_free(&buf);
 715
 716         /* Write a fake trailer so the pack functions play ball */
 717
 718         if ((error = append_to_pack(idx, &foo, GIT_OID_RAWSZ)) < 0)
 719                 goto cleanup;
 720
 721         idx->pack->mwf.size += GIT_OID_RAWSZ;
 722
 723         pentry = git__calloc(1, sizeof(struct git_pack_entry));
 724         GITERR_CHECK_ALLOC(pentry);
 725
 726         git_oid_cpy(&pentry->sha1, id);
 727         git_oid_cpy(&entry->oid, id);
 728         idx->off = entry_start + hdr_len + len;
 729
 730         error = save_entry(idx, entry, pentry, entry_start);
 731
 732 cleanup:
 733         if (error) {
 734                 git__free(entry);
 735                 git__free(pentry);
 736         }
 737
 738         git_odb_object_free(obj);
 739         return error;
 740 }
 741
 742 static int fix_thin_pack(git_indexer *idx, git_transfer_progress *stats)
 743 {
 744         int error, found_ref_delta = 0;
 745         unsigned int i;
 746         struct delta_info *delta;
 747         size_t size;
 748         git_otype type;
 749         git_mwindow *w = NULL;
 750         git_off_t curpos = 0;
 751         unsigned char *base_info;
 752         unsigned int left = 0;
 753         git_oid base;
 754
 755         assert(git_vector_length(&idx->deltas) > 0);
 756
 757         if (idx->odb == NULL) {
 758                 giterr_set(GITERR_INDEXER, "cannot fix a thin pack without an ODB");
 759                 return -1;
 760         }
 761
 762         /* Loop until we find the first REF delta */
 763         git_vector_foreach(&idx->deltas, i, delta) {
 764                 if (!delta)
 765                         continue;
 766
 767                 curpos = delta->delta_off;
 768                 error = git_packfile_unpack_header(&size, &type, &idx->pack->mwf, &w, &curpos);
 769                 git_mwindow_close(&w);
 770                 if (error < 0)
 771                         return error;
 772
 773                 if (type == GIT_OBJ_REF_DELTA) {
 774                         found_ref_delta = 1;
 775                         break;
 776                 }
 777         }
 778
 779         if (!found_ref_delta) {
 780                 giterr_set(GITERR_INDEXER, "no REF_DELTA found, cannot inject object");
 781                 return -1;
 782         }
 783
 784         /* curpos now points to the base information, which is an OID */
 785         base_info = git_mwindow_open(&idx->pack->mwf, &w, curpos, GIT_OID_RAWSZ, &left);
 786         if (base_info == NULL) {
 787                 giterr_set(GITERR_INDEXER, "failed to map delta information");
 788                 return -1;
 789         }
 790
 791         git_oid_fromraw(&base, base_info);
 792         git_mwindow_close(&w);
 793
 794         if (inject_object(idx, &base) < 0)
 795                 return -1;
 796
 797         stats->local_objects++;
 798
 799         return 0;
 800 }
 801
 802 static int resolve_deltas(git_indexer *idx, git_transfer_progress *stats)
 803 {
 804         unsigned int i;
 805         struct delta_info *delta;
 806         int progressed = 0, non_null = 0, progress_cb_result;
 807
 808         while (idx->deltas.length > 0) {
 809                 progressed = 0;
 810                 non_null = 0;
 811                 git_vector_foreach(&idx->deltas, i, delta) {
 812                         git_rawobj obj;
 813
 814                         if (!delta)
 815                                 continue;
 816
 817                         non_null = 1;
 818                         idx->off = delta->delta_off;
 819                         if (git_packfile_unpack(&obj, idx->pack, &idx->off) < 0)
 820                                 continue;
 821
 822                         if (hash_and_save(idx, &obj, delta->delta_off) < 0)
 823                                 continue;
 824
 825                         git__free(obj.data);
 826                         stats->indexed_objects++;
 827                         stats->indexed_deltas++;
 828                         progressed = 1;
 829                         if ((progress_cb_result = do_progress_callback(idx, stats)) < 0)
 830                                 return progress_cb_result;
 831
 832                         /* remove from the list */
 833                         git_vector_set(NULL, &idx->deltas, i, NULL);
 834                         git__free(delta);
 835                 }
 836
 837                 /* if none were actually set, we're done */
 838                 if (!non_null)
 839                         break;
 840
 841                 if (!progressed && (fix_thin_pack(idx, stats) < 0)) {
 842                         return -1;
 843                 }
 844         }
 845
 846         return 0;
 847 }
 848
 849 static int update_header_and_rehash(git_indexer *idx, git_transfer_progress *stats)
 850 {
 851         void *ptr;
 852         size_t chunk = 1024*1024;
 853         git_off_t hashed = 0;
 854         git_mwindow *w = NULL;
 855         git_mwindow_file *mwf;
 856         unsigned int left;
 857
 858         mwf = &idx->pack->mwf;
 859
 860         git_hash_init(&idx->trailer);
 861
 862
 863         /* Update the header to include the numer of local objects we injected */
 864         idx->hdr.hdr_entries = htonl(stats->total_objects + stats->local_objects);
 865         if (write_at(idx, &idx->hdr, 0, sizeof(struct git_pack_header)) < 0)
 866                 return -1;
 867
 868         /*
 869          * We now use the same technique as before to determine the
 870          * hash. We keep reading up to the end and let
 871          * hash_partially() keep the existing trailer out of the
 872          * calculation.
 873          */
 874         git_mwindow_free_all(mwf);
 875         idx->inbuf_len = 0;
 876         while (hashed < mwf->size) {
 877                 ptr = git_mwindow_open(mwf, &w, hashed, chunk, &left);
 878                 if (ptr == NULL)
 879                         return -1;
 880
 881                 hash_partially(idx, ptr, left);
 882                 hashed += left;
 883
 884                 git_mwindow_close(&w);
 885         }
 886
 887         return 0;
 888 }
 889
 890 int git_indexer_commit(git_indexer *idx, git_transfer_progress *stats)
 891 {
 892         git_mwindow *w = NULL;
 893         unsigned int i, long_offsets = 0, left;
 894         int error;
 895         struct git_pack_idx_header hdr;
 896         git_buf filename = GIT_BUF_INIT;
 897         struct entry *entry;
 898         git_oid trailer_hash, file_hash;
 899         git_hash_ctx ctx;
 900         git_filebuf index_file = {0};
 901         void *packfile_trailer;
 902
 903         if (git_hash_ctx_init(&ctx) < 0)
 904                 return -1;
 905
 906         /* Test for this before resolve_deltas(), as it plays with idx->off */
 907         if (idx->off < idx->pack->mwf.size - 20) {
 908                 giterr_set(GITERR_INDEXER, "Unexpected data at the end of the pack");
 909                 return -1;
 910         }
 911
 912         packfile_trailer = git_mwindow_open(&idx->pack->mwf, &w, idx->pack->mwf.size - GIT_OID_RAWSZ, GIT_OID_RAWSZ, &left);
 913         if (packfile_trailer == NULL) {
 914                 git_mwindow_close(&w);
 915                 goto on_error;
 916         }
 917
 918         /* Compare the packfile trailer as it was sent to us and what we calculated */
 919         git_oid_fromraw(&file_hash, packfile_trailer);
 920         git_mwindow_close(&w);
 921
 922         git_hash_final(&trailer_hash, &idx->trailer);
 923         if (git_oid_cmp(&file_hash, &trailer_hash)) {
 924                 giterr_set(GITERR_INDEXER, "packfile trailer mismatch");
 925                 return -1;
 926         }
 927
 928         /* Freeze the number of deltas */
 929         stats->total_deltas = stats->total_objects - stats->indexed_objects;
 930
 931         if ((error = resolve_deltas(idx, stats)) < 0)
 932                 return error;
 933
 934         if (stats->indexed_objects != stats->total_objects) {
 935                 giterr_set(GITERR_INDEXER, "early EOF");
 936                 return -1;
 937         }
 938
 939         if (stats->local_objects > 0) {
 940                 if (update_header_and_rehash(idx, stats) < 0)
 941                         return -1;
 942
 943                 git_hash_final(&trailer_hash, &idx->trailer);
 944                 write_at(idx, &trailer_hash, idx->pack->mwf.size - GIT_OID_RAWSZ, GIT_OID_RAWSZ);
 945         }
 946
 947         git_vector_sort(&idx->objects);
 948
 949         git_buf_sets(&filename, idx->pack->pack_name);
 950         git_buf_shorten(&filename, strlen("pack"));
 951         git_buf_puts(&filename, "idx");
 952         if (git_buf_oom(&filename))
 953                 return -1;
 954
 955         if (git_filebuf_open(&index_file, filename.ptr,
 956                 GIT_FILEBUF_HASH_CONTENTS, idx->mode) < 0)
 957                 goto on_error;
 958
 959         /* Write out the header */
 960         hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
 961         hdr.idx_version = htonl(2);
 962         git_filebuf_write(&index_file, &hdr, sizeof(hdr));
 963
 964         /* Write out the fanout table */
 965         for (i = 0; i < 256; ++i) {
 966                 uint32_t n = htonl(idx->fanout[i]);
 967                 git_filebuf_write(&index_file, &n, sizeof(n));
 968         }
 969
 970         /* Write out the object names (SHA-1 hashes) */
 971         git_vector_foreach(&idx->objects, i, entry) {
 972                 git_filebuf_write(&index_file, &entry->oid, sizeof(git_oid));
 973                 git_hash_update(&ctx, &entry->oid, GIT_OID_RAWSZ);
 974         }
 975         git_hash_final(&idx->hash, &ctx);
 976
 977         /* Write out the CRC32 values */
 978         git_vector_foreach(&idx->objects, i, entry) {
 979                 git_filebuf_write(&index_file, &entry->crc, sizeof(uint32_t));
 980         }
 981
 982         /* Write out the offsets */
 983         git_vector_foreach(&idx->objects, i, entry) {
 984                 uint32_t n;
 985
 986                 if (entry->offset == UINT32_MAX)
 987                         n = htonl(0x80000000 | long_offsets++);
 988                 else
 989                         n = htonl(entry->offset);
 990
 991                 git_filebuf_write(&index_file, &n, sizeof(uint32_t));
 992         }
 993
 994         /* Write out the long offsets */
 995         git_vector_foreach(&idx->objects, i, entry) {
 996                 uint32_t split[2];
 997
 998                 if (entry->offset != UINT32_MAX)
 999                         continue;
1000
1001                 split[0] = htonl(entry->offset_long >> 32);
1002                 split[1] = htonl(entry->offset_long & 0xffffffff);
1003
1004                 git_filebuf_write(&index_file, &split, sizeof(uint32_t) * 2);
1005         }
1006
1007         /* Write out the packfile trailer to the index */
1008         if (git_filebuf_write(&index_file, &trailer_hash, GIT_OID_RAWSZ) < 0)
1009                 goto on_error;
1010
1011         /* Write out the hash of the idx */
1012         if (git_filebuf_hash(&trailer_hash, &index_file) < 0)
1013                 goto on_error;
1014
1015         git_filebuf_write(&index_file, &trailer_hash, sizeof(git_oid));
1016
1017         /* Figure out what the final name should be */
1018         if (index_path(&filename, idx, ".idx") < 0)
1019                 goto on_error;
1020
1021         /* Commit file */
1022         if (git_filebuf_commit_at(&index_file, filename.ptr) < 0)
1023                 goto on_error;
1024
1025         git_mwindow_free_all(&idx->pack->mwf);
1026         /* We need to close the descriptor here so Windows doesn't choke on commit_at */
1027         if (p_close(idx->pack->mwf.fd) < 0) {
1028                 giterr_set(GITERR_OS, "failed to close packfile");
1029                 goto on_error;
1030         }
1031
1032         idx->pack->mwf.fd = -1;
1033
1034         if (index_path(&filename, idx, ".pack") < 0)
1035                 goto on_error;
1036
1037         /* And don't forget to rename the packfile to its new place. */
1038         p_rename(idx->pack->pack_name, git_buf_cstr(&filename));
1039
1040         git_buf_free(&filename);
1041         git_hash_ctx_cleanup(&ctx);
1042         return 0;
1043
1044 on_error:
1045         git_mwindow_free_all(&idx->pack->mwf);
1046         git_filebuf_cleanup(&index_file);
1047         git_buf_free(&filename);
1048         git_hash_ctx_cleanup(&ctx);
1049         return -1;
1050 }
1051
1052 void git_indexer_free(git_indexer *idx)
1053 {
1054         if (idx == NULL)
1055                 return;
1056
1057         git_vector_free_deep(&idx->objects);
1058
1059         if (idx->pack && idx->pack->idx_cache) {
1060                 struct git_pack_entry *pentry;
1061                 kh_foreach_value(
1062                         idx->pack->idx_cache, pentry, { git__free(pentry); });
1063
1064                 git_oidmap_free(idx->pack->idx_cache);
1065         }
1066
1067         git_vector_free_deep(&idx->deltas);
1068
1069         if (!git_mutex_lock(&git__mwindow_mutex)) {
1070                 git_packfile_free(idx->pack);
1071                 git_mutex_unlock(&git__mwindow_mutex);
1072         }
1073
1074         git_hash_ctx_cleanup(&idx->trailer);
1075         git_hash_ctx_cleanup(&idx->hash_ctx);
1076         git__free(idx);
1077 }