indexer.c 24.8 KB
Newer Older
1
/*
Edward Thomson committed
2
 * Copyright (C) the libgit2 contributors. All rights reserved.
3
 *
Vicent Marti committed
4 5
 * This file is part of libgit2, distributed under the GNU GPL v2 with
 * a Linking Exception. For full terms see the included COPYING file.
6 7
 */

Carlos Martín Nieto committed
8
#include "git2/indexer.h"
9
#include "git2/object.h"
Carlos Martín Nieto committed
10

11 12
#include "common.h"
#include "pack.h"
Carlos Martín Nieto committed
13
#include "mwindow.h"
14
#include "posix.h"
15 16
#include "pack.h"
#include "filebuf.h"
17
#include "oid.h"
18
#include "oidmap.h"
19
#include "zstream.h"
20

21 22
GIT__USE_OIDMAP;

23 24
extern git_mutex git__mwindow_mutex;

25
#define UINT31_MAX (0x7FFFFFFF)
26

27
struct entry {
28
	git_oid oid;
29 30 31 32 33
	uint32_t crc;
	uint32_t offset;
	uint64_t offset_long;
};

34
struct git_indexer {
35
	unsigned int parsed_header :1,
36
		opened_pack :1,
37 38
		have_stream :1,
		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 44
	git_off_t entry_start;
	git_packfile_stream stream;
45 46 47 48
	size_t nr_objects;
	git_vector objects;
	git_vector deltas;
	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 56 57
	/* Needed to look up objects which we want to inject to fix a thin pack */
	git_odb *odb;

58 59
	/* Fields for calculating the packfile trailer (hash of everything before it) */
	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 71 72
{
	return &idx->hash;
}

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 82 83 84
	memcpy(hdr, map.data, sizeof(*hdr));
	p_munmap(&map);

	/* Verify we recognize this pack file format. */
85
	if (hdr->hdr_signature != ntohl(PACK_SIGNATURE)) {
86
		giterr_set(GITERR_INDEXER, "Wrong pack signature");
87 88
		return -1;
	}
89

90
	if (!pack_version_ok(hdr->hdr_version)) {
91
		giterr_set(GITERR_INDEXER, "Wrong pack version");
92 93
		return -1;
	}
Carlos Martín Nieto committed
94

95
	return 0;
96 97
}

98
static int objects_cmp(const void *a, const void *b)
99 100 101 102
{
	const struct entry *entrya = a;
	const struct entry *entryb = b;

103
	return git_oid__cmp(&entrya->oid, &entryb->oid);
104 105
}

106 107
int git_indexer_new(
		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;
122 123
	idx->progress_cb = progress_cb;
	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);
127 128 129 130 131

	error = git_buf_joinpath(&path, prefix, suff);
	if (error < 0)
		goto cleanup;

132
	fd = git_futils_mktmp(&tmp_path, git_buf_cstr(&path), idx->mode);
133
	git_buf_free(&path);
134 135 136 137 138 139
	if (fd < 0)
		goto cleanup;

	error = git_packfile_alloc(&idx->pack, git_buf_cstr(&tmp_path));
	git_buf_free(&tmp_path);

140 141 142
	if (error < 0)
		goto cleanup;

143 144 145 146
	idx->pack->mwf.fd = fd;
	if ((error = git_mwindow_file_register(&idx->pack->mwf)) < 0)
		goto cleanup;

147 148 149 150
	*out = idx;
	return 0;

cleanup:
151 152 153
	if (fd != -1)
		p_close(fd);

154
	git_buf_free(&path);
155
	git_buf_free(&tmp_path);
156 157 158 159 160
	git__free(idx);
	return -1;
}

/* Try to store the delta so we can try to resolve it later */
161
static int store_delta(git_indexer *idx)
162
{
163 164
	struct delta_info *delta;

165 166
	delta = git__calloc(1, sizeof(struct delta_info));
	GITERR_CHECK_ALLOC(delta);
167
	delta->delta_off = idx->entry_start;
168

169
	if (git_vector_insert(&idx->deltas, delta) < 0)
170 171 172 173 174
		return -1;

	return 0;
}

175 176 177 178 179
static void hash_header(git_hash_ctx *ctx, git_off_t len, git_otype type)
{
	char buffer[64];
	size_t hdrlen;

180
	hdrlen = git_odb__format_object_header(buffer, sizeof(buffer), (size_t)len, type);
181 182 183
	git_hash_update(ctx, buffer, hdrlen);
}

184
static int hash_object_stream(git_indexer*idx, git_packfile_stream *stream)
185 186 187
{
	ssize_t read;

188
	assert(idx && stream);
189 190

	do {
191
		if ((read = git_packfile_stream_read(stream, idx->objbuf, sizeof(idx->objbuf))) < 0)
192 193
			break;

194
		git_hash_update(&idx->hash_ctx, idx->objbuf, read);
195 196 197 198 199 200 201 202
	} while (read > 0);

	if (read < 0)
		return (int)read;

	return 0;
}

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)
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222
{
	git_mwindow *w = NULL;

	assert(type == GIT_OBJ_REF_DELTA || type == GIT_OBJ_OFS_DELTA);

	if (type == GIT_OBJ_REF_DELTA) {
		idx->off += GIT_OID_RAWSZ;
	} else {
		git_off_t base_off = get_delta_base(idx->pack, &w, &idx->off, type, idx->entry_start);
		git_mwindow_close(&w);
		if (base_off < 0)
			return (int)base_off;
	}

	return 0;
}

/* Read from the stream and discard any output */
223
static int read_object_stream(git_indexer *idx, git_packfile_stream *stream)
224 225 226 227 228 229
{
	ssize_t read;

	assert(stream);

	do {
230
		read = git_packfile_stream_read(stream, idx->objbuf, sizeof(idx->objbuf));
231 232 233 234 235 236 237 238
	} while (read > 0);

	if (read < 0)
		return (int)read;

	return 0;
}

239 240 241 242 243 244 245 246 247
static int crc_object(uint32_t *crc_out, git_mwindow_file *mwf, git_off_t start, git_off_t size)
{
	void *ptr;
	uint32_t crc;
	unsigned int left, len;
	git_mwindow *w = NULL;

	crc = crc32(0L, Z_NULL, 0);
	while (size) {
248
		ptr = git_mwindow_open(mwf, &w, start, (size_t)size, &left);
249 250 251
		if (ptr == NULL)
			return -1;

252
		len = min(left, (unsigned int)size);
253 254 255 256 257 258 259 260 261 262
		crc = crc32(crc, ptr, len);
		size -= len;
		start += len;
		git_mwindow_close(&w);
	}

	*crc_out = htonl(crc);
	return 0;
}

263
static int store_object(git_indexer *idx)
264
{
265 266
	int i, error;
	khiter_t k;
267 268 269 270
	git_oid oid;
	struct entry *entry;
	git_off_t entry_size;
	struct git_pack_entry *pentry;
271
	git_off_t entry_start = idx->entry_start;
272 273 274 275

	entry = git__calloc(1, sizeof(*entry));
	GITERR_CHECK_ALLOC(entry);

Linquize committed
276
	pentry = git__calloc(1, sizeof(struct git_pack_entry));
277 278
	GITERR_CHECK_ALLOC(pentry);

279
	git_hash_final(&oid, &idx->hash_ctx);
280 281 282 283 284 285 286 287 288 289
	entry_size = idx->off - entry_start;
	if (entry_start > UINT31_MAX) {
		entry->offset = UINT32_MAX;
		entry->offset_long = entry_start;
	} else {
		entry->offset = (uint32_t)entry_start;
	}

	git_oid_cpy(&pentry->sha1, &oid);
	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

298 299 300 301 302 303 304
	if (error == 0) {
		giterr_set(GITERR_INDEXER, "duplicate object %s found in pack", git_oid_tostr_s(&pentry->sha1));
		git__free(pentry);
		goto on_error;
	}


305 306
	kh_value(idx->pack->idx_cache, k) = pentry;

307 308
	git_oid_cpy(&entry->oid, &oid);

309
	if (crc_object(&entry->crc, &idx->pack->mwf, entry_start, entry_size) < 0)
310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
		goto on_error;

	/* Add the object to the list */
	if (git_vector_insert(&idx->objects, entry) < 0)
		goto on_error;

	for (i = oid.id[0]; i < 256; ++i) {
		idx->fanout[i]++;
	}

	return 0;

on_error:
	git__free(entry);

	return -1;
}

328
static int save_entry(git_indexer *idx, struct entry *entry, struct git_pack_entry *pentry, git_off_t entry_start)
329
{
330 331
	int i, error;
	khiter_t k;
332 333 334 335 336 337 338 339

	if (entry_start > UINT31_MAX) {
		entry->offset = UINT32_MAX;
		entry->offset_long = entry_start;
	} else {
		entry->offset = (uint32_t)entry_start;
	}

340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357
	pentry->offset = entry_start;
	k = kh_put(oid, idx->pack->idx_cache, &pentry->sha1, &error);
	if (!error)
		return -1;

	kh_value(idx->pack->idx_cache, k) = pentry;

	/* Add the object to the list */
	if (git_vector_insert(&idx->objects, entry) < 0)
		return -1;

	for (i = entry->oid.id[0]; i < 256; ++i) {
		idx->fanout[i]++;
	}

	return 0;
}

358
static int hash_and_save(git_indexer *idx, git_rawobj *obj, git_off_t entry_start)
359 360 361 362
{
	git_oid oid;
	size_t entry_size;
	struct entry *entry;
363
	struct git_pack_entry *pentry = NULL;
364 365 366 367

	entry = git__calloc(1, sizeof(*entry));
	GITERR_CHECK_ALLOC(entry);

368
	if (git_odb__hashobj(&oid, obj) < 0) {
369
		giterr_set(GITERR_INDEXER, "Failed to hash object");
370
		goto on_error;
371 372
	}

Linquize committed
373
	pentry = git__calloc(1, sizeof(struct git_pack_entry));
374 375 376 377 378 379 380
	GITERR_CHECK_ALLOC(pentry);

	git_oid_cpy(&pentry->sha1, &oid);
	git_oid_cpy(&entry->oid, &oid);
	entry->crc = crc32(0L, Z_NULL, 0);

	entry_size = (size_t)(idx->off - entry_start);
381
	if (crc_object(&entry->crc, &idx->pack->mwf, entry_start, entry_size) < 0)
382 383
		goto on_error;

384
	return save_entry(idx, entry, pentry, entry_start);
385

386
on_error:
387
	git__free(pentry);
388 389
	git__free(entry);
	git__free(obj->data);
390 391
	return -1;
}
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(
397 398
			idx->progress_cb(stats, idx->progress_payload),
			"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;
406 407 408 409 410

	if (size == 0)
		return;

	/* Easy case, dump the buffer and the data minus the last 20 bytes */
411
	if (size >= GIT_OID_RAWSZ) {
412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
		git_hash_update(&idx->trailer, idx->inbuf, idx->inbuf_len);
		git_hash_update(&idx->trailer, data, size - GIT_OID_RAWSZ);

		data += size - GIT_OID_RAWSZ;
		memcpy(idx->inbuf, data, GIT_OID_RAWSZ);
		idx->inbuf_len = GIT_OID_RAWSZ;
		return;
	}

	/* We can just append */
	if (idx->inbuf_len + size <= GIT_OID_RAWSZ) {
		memcpy(idx->inbuf + idx->inbuf_len, data, size);
		idx->inbuf_len += size;
		return;
	}

	/* We need to partially drain the buffer and then append */
429 430
	to_keep   = GIT_OID_RAWSZ - size;
	to_expell = idx->inbuf_len - to_keep;
431 432 433 434 435 436 437 438

	git_hash_update(&idx->trailer, idx->inbuf, to_expell);

	memmove(idx->inbuf, idx->inbuf + to_expell, to_keep);
	memcpy(idx->inbuf + to_keep, data, size);
	idx->inbuf_len += size - to_expell;
}

439 440 441
static int write_at(git_indexer *idx, const void *data, git_off_t offset, size_t size)
{
	git_file fd = idx->pack->mwf.fd;
442 443 444
	size_t page_size;
	size_t page_offset;
	git_off_t page_start;
445
	unsigned char *map_data;
446 447 448
	git_map map;
	int error;

449 450
	assert(data && size);

451 452 453
	if ((error = git__page_size(&page_size)) < 0)
		return error;

454
	/* the offset needs to be at the beginning of the a page boundary */
455 456
	page_offset = offset % page_size;
	page_start = offset - page_offset;
457 458 459 460

	if ((error = p_mmap(&map, page_offset + size, GIT_PROT_WRITE, GIT_MAP_SHARED, fd, page_start)) < 0)
		return error;

461 462
	map_data = (unsigned char *)map.data;
	memcpy(map_data + page_offset, data, size);
463 464 465 466 467 468 469 470 471
	p_munmap(&map);

	return 0;
}

static int append_to_pack(git_indexer *idx, const void *data, size_t size)
{
	git_off_t current_size = idx->pack->mwf.size;

472 473 474
	if (!size)
		return 0;

475 476
	/* add the extra space we need at the end */
	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);
478 479 480 481 482 483
		return -1;
	}

	return write_at(idx, data, idx->pack->mwf.size, size);
}

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

491 492
	assert(idx && data && stats);

493
	processed = stats->indexed_objects;
494

495
	if ((error = append_to_pack(idx, data, size)) < 0)
496
		return error;
497

nulltoken committed
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) {
504 505
		unsigned int total_objects;

506
		if ((unsigned)idx->pack->mwf.size < sizeof(struct git_pack_header))
507 508
			return 0;

509 510
		if ((error = parse_header(&idx->hdr, idx->pack)) < 0)
			return error;
511 512

		idx->parsed_header = 1;
513
		idx->nr_objects = ntohl(hdr->hdr_entries);
514 515 516 517
		idx->off = sizeof(struct git_pack_header);

		/* for now, limit to 2^32 objects */
		assert(idx->nr_objects == (size_t)((unsigned int)idx->nr_objects));
518 519 520 521
		if (idx->nr_objects == (size_t)((unsigned int)idx->nr_objects))
			total_objects = (unsigned int)idx->nr_objects;
		else
			total_objects = UINT_MAX;
522

523 524
		idx->pack->idx_cache = git_oidmap_alloc();
		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)
528 529
			return -1;

530
		if (git_vector_init(&idx->deltas, total_objects / 2, NULL) < 0)
531 532
			return -1;

533
		stats->received_objects = 0;
534
		stats->local_objects = 0;
535 536
		stats->total_deltas = 0;
		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;
542 543 544 545 546 547
	}

	/* Now that we have data in the pack, let's try to parse it */

	/* As the file grows any windows we try to use will be out of date */
	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;
552 553 554
		size_t entry_size;
		git_otype type;
		git_mwindow *w = NULL;
555

556 557 558
		if (idx->pack->mwf.size <= idx->off + 20)
			return 0;

559 560
		if (!idx->have_stream) {
			error = git_packfile_unpack_header(&entry_size, &type, mwf, &w, &idx->off);
561 562
			if (error == GIT_EBUFS) {
				idx->off = entry_start;
563
				return 0;
564
			}
565
			if (error < 0)
566
				goto on_error;
567 568 569

			git_mwindow_close(&w);
			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);
574 575 576 577 578
				if (error == GIT_EBUFS) {
					idx->off = entry_start;
					return 0;
				}
				if (error < 0)
579
					goto on_error;
580

581 582 583 584
				idx->have_delta = 1;
			} else {
				idx->have_delta = 0;
				hash_header(&idx->hash_ctx, entry_size, type);
585
			}
586

587
			idx->have_stream = 1;
588

589 590 591
			error = git_packfile_stream_open(stream, idx->pack, idx->off);
			if (error < 0)
				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;
603 604 605 606 607

		/* We want to free the stream reasorces no matter what here */
		idx->have_stream = 0;
		git_packfile_stream_free(stream);

608
		if (error < 0)
609 610
			goto on_error;

611 612 613 614 615 616 617
		if (idx->have_delta) {
			error = store_delta(idx);
		} else {
			error = store_object(idx);
		}

		if (error < 0)
618
			goto on_error;
619

620 621 622
		if (!idx->have_delta) {
			stats->indexed_objects = (unsigned int)++processed;
		}
623
		stats->received_objects++;
624

625
		if ((error = do_progress_callback(idx, stats)) != 0)
626
			goto on_error;
627
	}
628

629
	return 0;
630

631 632
on_error:
	git_mwindow_free_all(mwf);
633
	return error;
634
}
635

636
static int index_path(git_buf *path, git_indexer *idx, const char *suffix)
637 638 639
{
	const char prefix[] = "pack-";
	size_t slash = (size_t)path->size;
640

641 642 643
	/* search backwards for '/' */
	while (slash > 0 && path->ptr[slash - 1] != '/')
		slash--;
644

645 646 647 648 649 650
	if (git_buf_grow(path, slash + 1 + strlen(prefix) +
					 GIT_OID_HEXSZ + strlen(suffix) + 1) < 0)
		return -1;

	git_buf_truncate(path, slash);
	git_buf_puts(path, prefix);
nulltoken committed
651
	git_oid_fmt(path->ptr + git_buf_len(path), &idx->hash);
652 653 654 655 656 657
	path->size += GIT_OID_HEXSZ;
	git_buf_puts(path, suffix);

	return git_buf_oom(path) ? -1 : 0;
}

658 659 660 661
/**
 * Rewind the packfile by the trailer, as we might need to fix the
 * packfile by injecting objects at the tail and must overwrite it.
 */
662
static void seek_back_trailer(git_indexer *idx)
663 664 665 666 667
{
	idx->pack->mwf.size -= GIT_OID_RAWSZ;
	git_mwindow_free_all(&idx->pack->mwf);
}

668
static int inject_object(git_indexer *idx, git_oid *id)
669
{
670 671
	git_odb_object *obj;
	struct entry *entry;
672
	struct git_pack_entry *pentry = NULL;
673 674 675 676 677 678 679 680
	git_oid foo = {{0}};
	unsigned char hdr[64];
	git_buf buf = GIT_BUF_INIT;
	git_off_t entry_start;
	const void *data;
	size_t len, hdr_len;
	int error;

681 682
	seek_back_trailer(idx);
	entry_start = idx->pack->mwf.size;
683

684 685
	if (git_odb_read(&obj, idx->odb, id) < 0) {
		giterr_set(GITERR_INDEXER, "missing delta bases");
686
		return -1;
687
	}
688 689 690 691

	data = git_odb_object_data(obj);
	len = git_odb_object_size(obj);

692 693 694
	entry = git__calloc(1, sizeof(*entry));
	GITERR_CHECK_ALLOC(entry);

695 696 697 698
	entry->crc = crc32(0L, Z_NULL, 0);

	/* Write out the object header */
	hdr_len = git_packfile__object_header(hdr, len, git_odb_object_type(obj));
699 700 701
	if ((error = append_to_pack(idx, hdr, hdr_len)) < 0)
		goto cleanup;

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)
706 707 708
		goto cleanup;

	/* And then the compressed object */
709 710 711
	if ((error = append_to_pack(idx, buf.ptr, buf.size)) < 0)
		goto cleanup;

712
	idx->pack->mwf.size += buf.size;
Linquize committed
713
	entry->crc = htonl(crc32(entry->crc, (unsigned char *)buf.ptr, (uInt)buf.size));
714 715 716
	git_buf_free(&buf);

	/* Write a fake trailer so the pack functions play ball */
717 718

	if ((error = append_to_pack(idx, &foo, GIT_OID_RAWSZ)) < 0)
719 720 721 722 723 724 725 726 727 728 729
		goto cleanup;

	idx->pack->mwf.size += GIT_OID_RAWSZ;

	pentry = git__calloc(1, sizeof(struct git_pack_entry));
	GITERR_CHECK_ALLOC(pentry);

	git_oid_cpy(&pentry->sha1, id);
	git_oid_cpy(&entry->oid, id);
	idx->off = entry_start + hdr_len + len;

730
	error = save_entry(idx, entry, pentry, entry_start);
731 732

cleanup:
733 734 735 736
	if (error) {
		git__free(entry);
		git__free(pentry);
	}
737

738 739 740 741
	git_odb_object_free(obj);
	return error;
}

742
static int fix_thin_pack(git_indexer *idx, git_transfer_progress *stats)
743
{
744
	int error, found_ref_delta = 0;
745 746
	unsigned int i;
	struct delta_info *delta;
747 748 749
	size_t size;
	git_otype type;
	git_mwindow *w = NULL;
Linquize committed
750
	git_off_t curpos = 0;
751 752 753 754 755
	unsigned char *base_info;
	unsigned int left = 0;
	git_oid base;

	assert(git_vector_length(&idx->deltas) > 0);
756 757 758 759 760

	if (idx->odb == NULL) {
		giterr_set(GITERR_INDEXER, "cannot fix a thin pack without an ODB");
		return -1;
	}
761

762
	/* Loop until we find the first REF delta */
763
	git_vector_foreach(&idx->deltas, i, delta) {
764 765 766
		if (!delta)
			continue;

767
		curpos = delta->delta_off;
768 769 770 771 772
		error = git_packfile_unpack_header(&size, &type, &idx->pack->mwf, &w, &curpos);
		git_mwindow_close(&w);
		if (error < 0)
			return error;

773 774 775
		if (type == GIT_OBJ_REF_DELTA) {
			found_ref_delta = 1;
			break;
776
		}
777
	}
778

779 780 781 782
	if (!found_ref_delta) {
		giterr_set(GITERR_INDEXER, "no REF_DELTA found, cannot inject object");
		return -1;
	}
783

784 785 786 787 788 789
	/* curpos now points to the base information, which is an OID */
	base_info = git_mwindow_open(&idx->pack->mwf, &w, curpos, GIT_OID_RAWSZ, &left);
	if (base_info == NULL) {
		giterr_set(GITERR_INDEXER, "failed to map delta information");
		return -1;
	}
790

791 792
	git_oid_fromraw(&base, base_info);
	git_mwindow_close(&w);
793

794 795 796 797
	if (inject_object(idx, &base) < 0)
		return -1;

	stats->local_objects++;
798 799 800 801

	return 0;
}

802
static int resolve_deltas(git_indexer *idx, git_transfer_progress *stats)
803 804 805
{
	unsigned int i;
	struct delta_info *delta;
806
	int progressed = 0, non_null = 0, progress_cb_result;
807 808 809

	while (idx->deltas.length > 0) {
		progressed = 0;
810
		non_null = 0;
811 812 813
		git_vector_foreach(&idx->deltas, i, delta) {
			git_rawobj obj;

814 815 816 817
			if (!delta)
				continue;

			non_null = 1;
818 819 820 821 822 823 824 825 826
			idx->off = delta->delta_off;
			if (git_packfile_unpack(&obj, idx->pack, &idx->off) < 0)
				continue;

			if (hash_and_save(idx, &obj, delta->delta_off) < 0)
				continue;

			git__free(obj.data);
			stats->indexed_objects++;
827
			stats->indexed_deltas++;
828
			progressed = 1;
829 830
			if ((progress_cb_result = do_progress_callback(idx, stats)) < 0)
				return progress_cb_result;
831

832 833
			/* remove from the list */
			git_vector_set(NULL, &idx->deltas, i, NULL);
834
			git__free(delta);
835
		}
836

837 838 839 840
		/* if none were actually set, we're done */
		if (!non_null)
			break;

841
		if (!progressed && (fix_thin_pack(idx, stats) < 0)) {
842
			return -1;
843
		}
844 845 846 847 848
	}

	return 0;
}

849
static int update_header_and_rehash(git_indexer *idx, git_transfer_progress *stats)
850 851 852 853 854 855 856 857 858 859
{
	void *ptr;
	size_t chunk = 1024*1024;
	git_off_t hashed = 0;
	git_mwindow *w = NULL;
	git_mwindow_file *mwf;
	unsigned int left;

	mwf = &idx->pack->mwf;

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

868 869 870 871 872 873
	/*
	 * We now use the same technique as before to determine the
	 * hash. We keep reading up to the end and let
	 * hash_partially() keep the existing trailer out of the
	 * calculation.
	 */
874
	git_mwindow_free_all(mwf);
875 876 877 878
	idx->inbuf_len = 0;
	while (hashed < mwf->size) {
		ptr = git_mwindow_open(mwf, &w, hashed, chunk, &left);
		if (ptr == NULL)
879
			return -1;
880

881 882 883 884
		hash_partially(idx, ptr, left);
		hashed += left;

		git_mwindow_close(&w);
885
	}
886

887 888 889
	return 0;
}

890
int git_indexer_commit(git_indexer *idx, git_transfer_progress *stats)
891 892 893
{
	git_mwindow *w = NULL;
	unsigned int i, long_offsets = 0, left;
894
	int error;
895 896 897
	struct git_pack_idx_header hdr;
	git_buf filename = GIT_BUF_INIT;
	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

903 904
	if (git_hash_ctx_init(&ctx) < 0)
		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");
909 910 911
		return -1;
	}

912 913 914 915 916 917 918 919 920 921 922 923
	packfile_trailer = git_mwindow_open(&idx->pack->mwf, &w, idx->pack->mwf.size - GIT_OID_RAWSZ, GIT_OID_RAWSZ, &left);
	if (packfile_trailer == NULL) {
		git_mwindow_close(&w);
		goto on_error;
	}

	/* Compare the packfile trailer as it was sent to us and what we calculated */
	git_oid_fromraw(&file_hash, packfile_trailer);
	git_mwindow_close(&w);

	git_hash_final(&trailer_hash, &idx->trailer);
	if (git_oid_cmp(&file_hash, &trailer_hash)) {
924
		giterr_set(GITERR_INDEXER, "packfile trailer mismatch");
925 926 927
		return -1;
	}

928 929 930
	/* Freeze the number of deltas */
	stats->total_deltas = stats->total_objects - stats->indexed_objects;

931 932
	if ((error = resolve_deltas(idx, stats)) < 0)
		return error;
933

934
	if (stats->indexed_objects != stats->total_objects) {
935
		giterr_set(GITERR_INDEXER, "early EOF");
936 937 938
		return -1;
	}

939 940 941 942 943
	if (stats->local_objects > 0) {
		if (update_header_and_rehash(idx, stats) < 0)
			return -1;

		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
	}

947 948 949
	git_vector_sort(&idx->objects);

	git_buf_sets(&filename, idx->pack->pack_name);
950
	git_buf_shorten(&filename, strlen("pack"));
951 952 953 954
	git_buf_puts(&filename, "idx");
	if (git_buf_oom(&filename))
		return -1;

955
	if (git_filebuf_open(&index_file, filename.ptr,
956
		GIT_FILEBUF_HASH_CONTENTS, idx->mode) < 0)
957 958 959 960 961
		goto on_error;

	/* Write out the header */
	hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
	hdr.idx_version = htonl(2);
962
	git_filebuf_write(&index_file, &hdr, sizeof(hdr));
963 964 965 966

	/* Write out the fanout table */
	for (i = 0; i < 256; ++i) {
		uint32_t n = htonl(idx->fanout[i]);
967
		git_filebuf_write(&index_file, &n, sizeof(n));
968 969
	}

970 971
	/* Write out the object names (SHA-1 hashes) */
	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);
976 977 978

	/* Write out the CRC32 values */
	git_vector_foreach(&idx->objects, i, entry) {
979
		git_filebuf_write(&index_file, &entry->crc, sizeof(uint32_t));
980 981 982 983 984 985 986 987 988 989 990
	}

	/* Write out the offsets */
	git_vector_foreach(&idx->objects, i, entry) {
		uint32_t n;

		if (entry->offset == UINT32_MAX)
			n = htonl(0x80000000 | long_offsets++);
		else
			n = htonl(entry->offset);

991
		git_filebuf_write(&index_file, &n, sizeof(uint32_t));
992 993 994 995 996 997 998 999 1000 1001 1002 1003
	}

	/* Write out the long offsets */
	git_vector_foreach(&idx->objects, i, entry) {
		uint32_t split[2];

		if (entry->offset != UINT32_MAX)
			continue;

		split[0] = htonl(entry->offset_long >> 32);
		split[1] = htonl(entry->offset_long & 0xffffffff);

1004
		git_filebuf_write(&index_file, &split, sizeof(uint32_t) * 2);
1005 1006
	}

1007 1008
	/* Write out the packfile trailer to the index */
	if (git_filebuf_write(&index_file, &trailer_hash, GIT_OID_RAWSZ) < 0)
1009 1010
		goto on_error;

1011 1012
	/* Write out the hash of the idx */
	if (git_filebuf_hash(&trailer_hash, &index_file) < 0)
1013 1014
		goto on_error;

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 1020 1021
		goto on_error;

	/* Commit file */
1022
	if (git_filebuf_commit_at(&index_file, filename.ptr) < 0)
1023 1024 1025
		goto on_error;

	git_mwindow_free_all(&idx->pack->mwf);
1026
	/* We need to close the descriptor here so Windows doesn't choke on commit_at */
1027 1028 1029 1030 1031
	if (p_close(idx->pack->mwf.fd) < 0) {
		giterr_set(GITERR_OS, "failed to close packfile");
		goto on_error;
	}

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 1043 1044
	return 0;

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 1050 1051
	return -1;
}

1052
void git_indexer_free(git_indexer *idx)
1053 1054 1055 1056
{
	if (idx == NULL)
		return;

1057
	git_vector_free_deep(&idx->objects);
1058

1059
	if (idx->pack && idx->pack->idx_cache) {
Russell Belfer committed
1060 1061 1062
		struct git_pack_entry *pentry;
		kh_foreach_value(
			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 1070 1071 1072 1073

	if (!git_mutex_lock(&git__mwindow_mutex)) {
		git_packfile_free(idx->pack);
		git_mutex_unlock(&git__mwindow_mutex);
	}

1074 1075
	git_hash_ctx_cleanup(&idx->trailer);
	git_hash_ctx_cleanup(&idx->hash_ctx);
1076 1077
	git__free(idx);
}