sha1_file.c 63.3 KB
Newer Older
1 2 3 4 5 6 7 8 9
/*
 * GIT - The information manager from hell
 *
 * Copyright (C) Linus Torvalds, 2005
 *
 * This handles basic git sha1 object files - packing, unpacking,
 * creation etc.
 */
#include "cache.h"
10
#include "delta.h"
11
#include "pack.h"
12 13 14 15
#include "blob.h"
#include "commit.h"
#include "tag.h"
#include "tree.h"
16
#include "refs.h"
17
#include "pack-revindex.h"
18
#include "sha1-lookup.h"
19

20 21 22 23 24 25 26 27
#ifndef O_NOATIME
#if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
#define O_NOATIME 01000000
#else
#define O_NOATIME 0
#endif
#endif

28 29
#ifdef NO_C99_FORMAT
#define SZ_FMT "lu"
30
static unsigned long sz_fmt(size_t s) { return (unsigned long)s; }
31 32
#else
#define SZ_FMT "zu"
33
static size_t sz_fmt(size_t s) { return s; }
34 35
#endif

36
const unsigned char null_sha1[20];
J
Junio C Hamano 已提交
37

38 39
int safe_create_leading_directories(char *path)
{
40
	char *pos = path + offset_1st_component(path);
41 42
	struct stat st;

43 44 45 46
	while (pos) {
		pos = strchr(pos, '/');
		if (!pos)
			break;
47 48 49 50 51
		while (*++pos == '/')
			;
		if (!*pos)
			break;
		*--pos = '\0';
52 53 54
		if (!stat(path, &st)) {
			/* path exists */
			if (!S_ISDIR(st.st_mode)) {
55
				*pos = '/';
56
				return -3;
57
			}
58
		}
59 60 61 62
		else if (mkdir(path, 0777)) {
			*pos = '/';
			return -1;
		}
63 64 65 66
		else if (adjust_shared_perm(path)) {
			*pos = '/';
			return -2;
		}
67 68 69 70
		*pos++ = '/';
	}
	return 0;
}
71

72 73 74 75 76 77 78 79 80
int safe_create_leading_directories_const(const char *path)
{
	/* path points to cache entries, so xstrdup before messing with it */
	char *buf = xstrdup(path);
	int result = safe_create_leading_directories(buf);
	free(buf);
	return result;
}

81 82 83 84 85 86 87 88 89 90 91 92
static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
{
	int i;
	for (i = 0; i < 20; i++) {
		static char hex[] = "0123456789abcdef";
		unsigned int val = sha1[i];
		char *pos = pathbuf + i*2 + (i > 0);
		*pos++ = hex[val >> 4];
		*pos = hex[val & 0xf];
	}
}

93 94
/*
 * NOTE! This returns a statically allocated buffer, so you have to be
95
 * careful about using it. Do an "xstrdup()" if you need to save the
96
 * filename.
97 98 99
 *
 * Also note that this returns the location for creating.  Reading
 * SHA1 file can happen from any alternate directory listed in the
J
Junio C Hamano 已提交
100
 * DB_ENVIRONMENT environment variable if it is not found in
101
 * the primary object database.
102 103 104 105 106 107
 */
char *sha1_file_name(const unsigned char *sha1)
{
	static char *name, *base;

	if (!base) {
J
Junio C Hamano 已提交
108
		const char *sha1_file_directory = get_object_directory();
109
		int len = strlen(sha1_file_directory);
110
		base = xmalloc(len + 60);
111 112 113 114 115 116
		memcpy(base, sha1_file_directory, len);
		memset(base+len, 0, 60);
		base[len] = '/';
		base[len+3] = '/';
		name = base + len + 1;
	}
117
	fill_sha1_path(name, sha1);
118 119 120
	return base;
}

121
static char *sha1_get_pack_name(const unsigned char *sha1,
J
Junio C Hamano 已提交
122
				char **name, char **base, const char *which)
123 124
{
	static const char hex[] = "0123456789abcdef";
125
	char *buf;
126 127
	int i;

128
	if (!*base) {
129 130
		const char *sha1_file_directory = get_object_directory();
		int len = strlen(sha1_file_directory);
131
		*base = xmalloc(len + 60);
J
Junio C Hamano 已提交
132 133
		sprintf(*base, "%s/pack/pack-1234567890123456789012345678901234567890.%s",
			sha1_file_directory, which);
134
		*name = *base + len + 11;
135 136
	}

137
	buf = *name;
138 139 140 141 142 143

	for (i = 0; i < 20; i++) {
		unsigned int val = *sha1++;
		*buf++ = hex[val >> 4];
		*buf++ = hex[val & 0xf];
	}
J
Junio C Hamano 已提交
144

145
	return *base;
146 147
}

148
char *sha1_pack_name(const unsigned char *sha1)
149
{
150
	static char *name, *base;
151

J
Junio C Hamano 已提交
152
	return sha1_get_pack_name(sha1, &name, &base, "pack");
153
}
154

155 156 157
char *sha1_pack_index_name(const unsigned char *sha1)
{
	static char *name, *base;
J
Junio C Hamano 已提交
158

J
Junio C Hamano 已提交
159
	return sha1_get_pack_name(sha1, &name, &base, "idx");
160 161
}

162 163
struct alternate_object_database *alt_odb_list;
static struct alternate_object_database **alt_odb_tail;
164

M
Martin Waitz 已提交
165 166
static void read_info_alternates(const char * alternates, int depth);

J
Junio C Hamano 已提交
167 168
/*
 * Prepare alternate object database registry.
169 170 171 172 173
 *
 * The variable alt_odb_list points at the list of struct
 * alternate_object_database.  The elements on this list come from
 * non-empty elements from colon separated ALTERNATE_DB_ENVIRONMENT
 * environment variable, and $GIT_OBJECT_DIRECTORY/info/alternates,
174 175
 * whose contents is similar to that environment variable but can be
 * LF separated.  Its base points at a statically allocated buffer that
176 177 178 179 180
 * contains "/the/directory/corresponding/to/.git/objects/...", while
 * its name points just after the slash at the end of ".git/objects/"
 * in the example above, and has enough space to hold 40-byte hex
 * SHA1, an extra slash for the first level indirection, and the
 * terminating NUL.
J
Junio C Hamano 已提交
181
 */
M
Martin Waitz 已提交
182
static int link_alt_odb_entry(const char * entry, int len, const char * relative_base, int depth)
183
{
184
	const char *objdir = get_object_directory();
M
Martin Waitz 已提交
185 186 187 188 189
	struct alternate_object_database *ent;
	struct alternate_object_database *alt;
	/* 43 = 40-byte + 2 '/' + terminating NUL */
	int pfxlen = len;
	int entlen = pfxlen + 43;
190
	int base_len = -1;
191

192
	if (!is_absolute_path(entry) && relative_base) {
M
Martin Waitz 已提交
193 194 195 196 197 198 199 200
		/* Relative alt-odb */
		if (base_len < 0)
			base_len = strlen(relative_base) + 1;
		entlen += base_len;
		pfxlen += base_len;
	}
	ent = xmalloc(sizeof(*ent) + entlen);

201
	if (!is_absolute_path(entry) && relative_base) {
M
Martin Waitz 已提交
202 203 204 205 206 207 208 209 210 211 212 213
		memcpy(ent->base, relative_base, base_len - 1);
		ent->base[base_len - 1] = '/';
		memcpy(ent->base + base_len, entry, len);
	}
	else
		memcpy(ent->base, entry, pfxlen);

	ent->name = ent->base + pfxlen + 1;
	ent->base[pfxlen + 3] = '/';
	ent->base[pfxlen] = ent->base[entlen-1] = 0;

	/* Detect cases where alternate disappeared */
214
	if (!is_directory(ent->base)) {
M
Martin Waitz 已提交
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
		error("object directory %s does not exist; "
		      "check .git/objects/info/alternates.",
		      ent->base);
		free(ent);
		return -1;
	}

	/* Prevent the common mistake of listing the same
	 * thing twice, or object directory itself.
	 */
	for (alt = alt_odb_list; alt; alt = alt->next) {
		if (!memcmp(ent->base, alt->base, pfxlen)) {
			free(ent);
			return -1;
		}
	}
	if (!memcmp(ent->base, objdir, pfxlen)) {
		free(ent);
		return -1;
	}

	/* add the alternate entry */
	*alt_odb_tail = ent;
	alt_odb_tail = &(ent->next);
	ent->next = NULL;

	/* recursively add alternates */
	read_info_alternates(ent->base, depth + 1);

	ent->base[pfxlen] = '/';

	return 0;
}

static void link_alt_odb_entries(const char *alt, const char *ep, int sep,
				 const char *relative_base, int depth)
{
	const char *cp, *last;

	if (depth > 5) {
		error("%s: ignoring alternate object stores, nesting too deep.",
				relative_base);
		return;
	}

260
	last = alt;
261 262 263 264 265 266 267 268
	while (last < ep) {
		cp = last;
		if (cp < ep && *cp == '#') {
			while (cp < ep && *cp != sep)
				cp++;
			last = cp + 1;
			continue;
		}
M
Martin Waitz 已提交
269 270
		while (cp < ep && *cp != sep)
			cp++;
271
		if (last != cp) {
272
			if (!is_absolute_path(last) && depth) {
M
Martin Waitz 已提交
273 274 275 276 277
				error("%s: ignoring relative alternate object store %s",
						relative_base, last);
			} else {
				link_alt_odb_entry(last, cp - last,
						relative_base, depth);
278
			}
279
		}
280
		while (cp < ep && *cp == sep)
281 282
			cp++;
		last = cp;
283
	}
284 285
}

M
Martin Waitz 已提交
286
static void read_info_alternates(const char * relative_base, int depth)
287
{
288
	char *map;
289
	size_t mapsz;
290
	struct stat st;
291 292 293 294 295
	const char alt_file_name[] = "info/alternates";
	/* Given that relative_base is no longer than PATH_MAX,
	   ensure that "path" has enough space to append "/", the
	   file name, "info/alternates", and a trailing NUL.  */
	char path[PATH_MAX + 1 + sizeof alt_file_name];
M
Martin Waitz 已提交
296
	int fd;
297

298
	sprintf(path, "%s/%s", relative_base, alt_file_name);
299 300 301 302 303
	fd = open(path, O_RDONLY);
	if (fd < 0)
		return;
	if (fstat(fd, &st) || (st.st_size == 0)) {
		close(fd);
304
		return;
305
	}
306 307
	mapsz = xsize_t(st.st_size);
	map = xmmap(NULL, mapsz, PROT_READ, MAP_PRIVATE, fd, 0);
308 309
	close(fd);

310
	link_alt_odb_entries(map, map + mapsz, '\n', relative_base, depth);
M
Martin Waitz 已提交
311

312
	munmap(map, mapsz);
313 314
}

315 316 317
void add_to_alternates_file(const char *reference)
{
	struct lock_file *lock = xcalloc(1, sizeof(struct lock_file));
318
	int fd = hold_lock_file_for_append(lock, git_path("objects/info/alternates"), LOCK_DIE_ON_ERROR);
319 320 321 322 323 324 325 326
	char *alt = mkpath("%s/objects\n", reference);
	write_or_die(fd, alt, strlen(alt));
	if (commit_lock_file(lock))
		die("could not close alternates file");
	if (alt_odb_tail)
		link_alt_odb_entries(alt, alt + strlen(alt), '\n', NULL, 0);
}

327 328 329 330 331 332 333 334 335 336
void foreach_alt_odb(alt_odb_fn fn, void *cb)
{
	struct alternate_object_database *ent;

	prepare_alt_odb();
	for (ent = alt_odb_list; ent; ent = ent->next)
		if (fn(ent, cb))
			return;
}

M
Martin Waitz 已提交
337 338
void prepare_alt_odb(void)
{
T
Timo Hirvonen 已提交
339
	const char *alt;
M
Martin Waitz 已提交
340

S
Shawn O. Pearce 已提交
341 342 343
	if (alt_odb_tail)
		return;

M
Martin Waitz 已提交
344 345 346 347
	alt = getenv(ALTERNATE_DB_ENVIRONMENT);
	if (!alt) alt = "";

	alt_odb_tail = &alt_odb_list;
348
	link_alt_odb_entries(alt, alt + strlen(alt), PATH_SEP, NULL, 0);
M
Martin Waitz 已提交
349 350 351 352

	read_info_alternates(get_object_directory(), 0);
}

353
static int has_loose_object_local(const unsigned char *sha1)
354 355
{
	char *name = sha1_file_name(sha1);
356 357
	return !access(name, F_OK);
}
358

359 360 361
int has_loose_object_nonlocal(const unsigned char *sha1)
{
	struct alternate_object_database *alt;
362
	prepare_alt_odb();
363
	for (alt = alt_odb_list; alt; alt = alt->next) {
364
		fill_sha1_path(alt->name, sha1);
365 366
		if (!access(alt->base, F_OK))
			return 1;
367
	}
368
	return 0;
369 370
}

371 372 373 374 375 376
static int has_loose_object(const unsigned char *sha1)
{
	return has_loose_object_local(sha1) ||
	       has_loose_object_nonlocal(sha1);
}

377
static unsigned int pack_used_ctr;
378 379 380 381
static unsigned int pack_mmap_calls;
static unsigned int peak_pack_open_windows;
static unsigned int pack_open_windows;
static size_t peak_pack_mapped;
382
static size_t pack_mapped;
383
struct packed_git *packed_git;
384

J
Junio C Hamano 已提交
385
void pack_report(void)
386 387
{
	fprintf(stderr,
388 389 390
		"pack_report: getpagesize()            = %10" SZ_FMT "\n"
		"pack_report: core.packedGitWindowSize = %10" SZ_FMT "\n"
		"pack_report: core.packedGitLimit      = %10" SZ_FMT "\n",
391 392 393
		sz_fmt(getpagesize()),
		sz_fmt(packed_git_window_size),
		sz_fmt(packed_git_limit));
394 395 396 397
	fprintf(stderr,
		"pack_report: pack_used_ctr            = %10u\n"
		"pack_report: pack_mmap_calls          = %10u\n"
		"pack_report: pack_open_windows        = %10u / %10u\n"
398 399
		"pack_report: pack_mapped              = "
			"%10" SZ_FMT " / %10" SZ_FMT "\n",
400 401 402
		pack_used_ctr,
		pack_mmap_calls,
		pack_open_windows, peak_pack_open_windows,
403
		sz_fmt(pack_mapped), sz_fmt(peak_pack_mapped));
404 405
}

406
static int check_packed_git_idx(const char *path,  struct packed_git *p)
407 408
{
	void *idx_map;
409
	struct pack_idx_header *hdr;
410
	size_t idx_size;
411
	uint32_t version, nr, i, *index;
412
	int fd = open(path, O_RDONLY);
413
	struct stat st;
414

415 416 417 418 419 420
	if (fd < 0)
		return -1;
	if (fstat(fd, &st)) {
		close(fd);
		return -1;
	}
421
	idx_size = xsize_t(st.st_size);
422 423 424 425
	if (idx_size < 4 * 256 + 20 + 20) {
		close(fd);
		return error("index file %s is too small", path);
	}
426
	idx_map = xmmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
427 428
	close(fd);

429 430
	hdr = idx_map;
	if (hdr->idx_signature == htonl(PACK_IDX_SIGNATURE)) {
431 432 433
		version = ntohl(hdr->idx_version);
		if (version < 2 || version > 2) {
			munmap(idx_map, idx_size);
434
			return error("index file %s is version %"PRIu32
435 436 437 438 439 440
				     " and is not supported by this binary"
				     " (try upgrading GIT to a newer version)",
				     path, version);
		}
	} else
		version = 1;
441

442
	nr = 0;
443
	index = idx_map;
444 445
	if (version > 1)
		index += 2;  /* skip index header */
446
	for (i = 0; i < 256; i++) {
447
		uint32_t n = ntohl(index[i]);
448 449
		if (n < nr) {
			munmap(idx_map, idx_size);
450
			return error("non-monotonic index %s", path);
451
		}
452 453 454
		nr = n;
	}

455 456 457 458 459 460 461 462 463 464
	if (version == 1) {
		/*
		 * Total size:
		 *  - 256 index entries 4 bytes each
		 *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
		 *  - 20-byte SHA1 of the packfile
		 *  - 20-byte SHA1 file checksum
		 */
		if (idx_size != 4*256 + nr * 24 + 20 + 20) {
			munmap(idx_map, idx_size);
465
			return error("wrong index v1 file size in %s", path);
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
		}
	} else if (version == 2) {
		/*
		 * Minimum size:
		 *  - 8 bytes of header
		 *  - 256 index entries 4 bytes each
		 *  - 20-byte sha1 entry * nr
		 *  - 4-byte crc entry * nr
		 *  - 4-byte offset entry * nr
		 *  - 20-byte SHA1 of the packfile
		 *  - 20-byte SHA1 file checksum
		 * And after the 4-byte offset table might be a
		 * variable sized table containing 8-byte entries
		 * for offsets larger than 2^31.
		 */
		unsigned long min_size = 8 + 4*256 + nr*(20 + 4 + 4) + 20 + 20;
L
Linus Torvalds 已提交
482 483 484 485
		unsigned long max_size = min_size;
		if (nr)
			max_size += (nr - 1)*8;
		if (idx_size < min_size || idx_size > max_size) {
486
			munmap(idx_map, idx_size);
487
			return error("wrong index v2 file size in %s", path);
488
		}
489 490 491 492 493 494 495 496 497
		if (idx_size != min_size &&
		    /*
		     * make sure we can deal with large pack offsets.
		     * 31-bit signed offset won't be enough, neither
		     * 32-bit unsigned one will be.
		     */
		    (sizeof(off_t) <= 4)) {
			munmap(idx_map, idx_size);
			return error("pack too large for current definition of off_t in %s", path);
498
		}
499
	}
500

501
	p->index_version = version;
502 503
	p->index_data = idx_map;
	p->index_size = idx_size;
N
Nicolas Pitre 已提交
504
	p->num_objects = nr;
505 506 507
	return 0;
}

508
int open_pack_index(struct packed_git *p)
509 510 511 512 513 514 515 516 517 518 519 520 521 522
{
	char *idx_name;
	int ret;

	if (p->index_data)
		return 0;

	idx_name = xstrdup(p->pack_name);
	strcpy(idx_name + strlen(idx_name) - strlen(".pack"), ".idx");
	ret = check_packed_git_idx(idx_name, p);
	free(idx_name);
	return ret;
}

523 524 525 526
static void scan_windows(struct packed_git *p,
	struct packed_git **lru_p,
	struct pack_window **lru_w,
	struct pack_window **lru_l)
527
{
528 529 530 531 532 533 534 535
	struct pack_window *w, *w_l;

	for (w_l = NULL, w = p->windows; w; w = w->next) {
		if (!w->inuse_cnt) {
			if (!*lru_w || w->last_used < (*lru_w)->last_used) {
				*lru_p = p;
				*lru_w = w;
				*lru_l = w_l;
536 537
			}
		}
538
		w_l = w;
539
	}
540 541
}

542
static int unuse_one_window(struct packed_git *current, int keep_fd)
543 544 545 546 547 548 549 550
{
	struct packed_git *p, *lru_p = NULL;
	struct pack_window *lru_w = NULL, *lru_l = NULL;

	if (current)
		scan_windows(current, &lru_p, &lru_w, &lru_l);
	for (p = packed_git; p; p = p->next)
		scan_windows(p, &lru_p, &lru_w, &lru_l);
551 552 553 554 555 556 557
	if (lru_p) {
		munmap(lru_w->base, lru_w->len);
		pack_mapped -= lru_w->len;
		if (lru_l)
			lru_l->next = lru_w->next;
		else {
			lru_p->windows = lru_w->next;
558
			if (!lru_p->windows && lru_p->pack_fd != keep_fd) {
559 560 561 562 563
				close(lru_p->pack_fd);
				lru_p->pack_fd = -1;
			}
		}
		free(lru_w);
564
		pack_open_windows--;
565 566 567
		return 1;
	}
	return 0;
568 569
}

570
void release_pack_memory(size_t need, int fd)
571 572
{
	size_t cur = pack_mapped;
573
	while (need >= (cur - pack_mapped) && unuse_one_window(NULL, fd))
574 575 576
		; /* nothing */
}

577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
void close_pack_windows(struct packed_git *p)
{
	while (p->windows) {
		struct pack_window *w = p->windows;

		if (w->inuse_cnt)
			die("pack '%s' still has open windows to it",
			    p->pack_name);
		munmap(w->base, w->len);
		pack_mapped -= w->len;
		pack_open_windows--;
		p->windows = w->next;
		free(w);
	}
}

593
void unuse_pack(struct pack_window **w_cursor)
594
{
595 596 597 598 599
	struct pack_window *w = *w_cursor;
	if (w) {
		w->inuse_cnt--;
		*w_cursor = NULL;
	}
600 601
}

602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618
/*
 * This is used by git-repack in case a newly created pack happens to
 * contain the same set of objects as an existing one.  In that case
 * the resulting file might be different even if its name would be the
 * same.  It is best to close any reference to the old pack before it is
 * replaced on disk.  Of course no index pointers nor windows for given pack
 * must subsist at this point.  If ever objects from this pack are requested
 * again, the new version of the pack will be reinitialized through
 * reprepare_packed_git().
 */
void free_pack_by_name(const char *pack_name)
{
	struct packed_git *p, **pp = &packed_git;

	while (*pp) {
		p = *pp;
		if (strcmp(pack_name, p->pack_name) == 0) {
619
			clear_delta_base_cache();
620 621 622 623 624 625 626 627 628 629 630 631 632 633
			close_pack_windows(p);
			if (p->pack_fd != -1)
				close(p->pack_fd);
			if (p->index_data)
				munmap((void *)p->index_data, p->index_size);
			free(p->bad_object_sha1);
			*pp = p->next;
			free(p);
			return;
		}
		pp = &p->next;
	}
}

634 635 636 637 638
/*
 * Do not call this directly as this leaks p->pack_fd on error return;
 * call open_packed_git() instead.
 */
static int open_packed_git_1(struct packed_git *p)
639
{
640 641 642 643
	struct stat st;
	struct pack_header hdr;
	unsigned char sha1[20];
	unsigned char *idx_sha1;
644
	long fd_flag;
645

646 647 648
	if (!p->index_data && open_pack_index(p))
		return error("packfile %s index unavailable", p->pack_name);

649
	p->pack_fd = open(p->pack_name, O_RDONLY);
650 651
	while (p->pack_fd < 0 && errno == EMFILE && unuse_one_window(p, -1))
		p->pack_fd = open(p->pack_name, O_RDONLY);
652
	if (p->pack_fd < 0 || fstat(p->pack_fd, &st))
653
		return -1;
654 655

	/* If we created the struct before we had the pack we lack size. */
656 657
	if (!p->pack_size) {
		if (!S_ISREG(st.st_mode))
658
			return error("packfile %s not a regular file", p->pack_name);
659
		p->pack_size = st.st_size;
660
	} else if (p->pack_size != st.st_size)
661
		return error("packfile %s size changed", p->pack_name);
662

663 664 665 666 667
	/* We leave these file descriptors open with sliding mmap;
	 * there is no point keeping them open across exec(), though.
	 */
	fd_flag = fcntl(p->pack_fd, F_GETFD, 0);
	if (fd_flag < 0)
668
		return error("cannot determine file descriptor flags");
669 670
	fd_flag |= FD_CLOEXEC;
	if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1)
671
		return error("cannot set FD_CLOEXEC");
672

673
	/* Verify we recognize this pack file format. */
674
	if (read_in_full(p->pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
675
		return error("file %s is far too short to be a packfile", p->pack_name);
676
	if (hdr.hdr_signature != htonl(PACK_SIGNATURE))
677
		return error("file %s is not a GIT packfile", p->pack_name);
678
	if (!pack_version_ok(hdr.hdr_version))
679 680
		return error("packfile %s is version %"PRIu32" and not"
			" supported (try upgrading GIT to a newer version)",
681 682 683
			p->pack_name, ntohl(hdr.hdr_version));

	/* Verify the pack matches its index. */
N
Nicolas Pitre 已提交
684
	if (p->num_objects != ntohl(hdr.hdr_entries))
685 686
		return error("packfile %s claims to have %"PRIu32" objects"
			     " while index indicates %"PRIu32" objects",
N
Nicolas Pitre 已提交
687 688
			     p->pack_name, ntohl(hdr.hdr_entries),
			     p->num_objects);
689
	if (lseek(p->pack_fd, p->pack_size - sizeof(sha1), SEEK_SET) == -1)
690
		return error("end of packfile %s is unavailable", p->pack_name);
691
	if (read_in_full(p->pack_fd, sha1, sizeof(sha1)) != sizeof(sha1))
692
		return error("packfile %s signature is unavailable", p->pack_name);
693
	idx_sha1 = ((unsigned char *)p->index_data) + p->index_size - 40;
694
	if (hashcmp(sha1, idx_sha1))
695 696
		return error("packfile %s does not match index", p->pack_name);
	return 0;
697 698
}

699 700 701 702 703 704 705 706 707 708 709
static int open_packed_git(struct packed_git *p)
{
	if (!open_packed_git_1(p))
		return 0;
	if (p->pack_fd != -1) {
		close(p->pack_fd);
		p->pack_fd = -1;
	}
	return -1;
}

710
static int in_window(struct pack_window *win, off_t offset)
711 712 713 714 715 716 717 718 719 720 721 722
{
	/* We must promise at least 20 bytes (one hash) after the
	 * offset is available from this window, otherwise the offset
	 * is not actually in this window and a different window (which
	 * has that one hash excess) must be used.  This is to support
	 * the object header and delta base parsing routines below.
	 */
	off_t win_off = win->offset;
	return win_off <= offset
		&& (offset + 20) <= (win_off + win->len);
}

723
unsigned char *use_pack(struct packed_git *p,
724
		struct pack_window **w_cursor,
725
		off_t offset,
726
		unsigned int *left)
727
{
728
	struct pack_window *win = *w_cursor;
729

730 731
	if (p->pack_fd == -1 && open_packed_git(p))
		die("packfile %s cannot be accessed", p->pack_name);
732

F
Felipe Contreras 已提交
733
	/* Since packfiles end in a hash of their content and it's
734 735 736 737 738 739 740 741 742 743 744 745 746 747 748
	 * pointless to ask for an offset into the middle of that
	 * hash, and the in_window function above wouldn't match
	 * don't allow an offset too close to the end of the file.
	 */
	if (offset > (p->pack_size - 20))
		die("offset beyond end of packfile (truncated pack?)");

	if (!win || !in_window(win, offset)) {
		if (win)
			win->inuse_cnt--;
		for (win = p->windows; win; win = win->next) {
			if (in_window(win, offset))
				break;
		}
		if (!win) {
749
			size_t window_align = packed_git_window_size / 2;
750
			off_t len;
751
			win = xcalloc(1, sizeof(*win));
752
			win->offset = (offset / window_align) * window_align;
753 754 755 756
			len = p->pack_size - win->offset;
			if (len > packed_git_window_size)
				len = packed_git_window_size;
			win->len = (size_t)len;
757
			pack_mapped += win->len;
758
			while (packed_git_limit < pack_mapped
759
				&& unuse_one_window(p, p->pack_fd))
760
				; /* nothing */
761
			win->base = xmmap(NULL, win->len,
762 763 764
				PROT_READ, MAP_PRIVATE,
				p->pack_fd, win->offset);
			if (win->base == MAP_FAILED)
765 766 767
				die("packfile %s cannot be mapped: %s",
					p->pack_name,
					strerror(errno));
768 769 770 771 772 773
			pack_mmap_calls++;
			pack_open_windows++;
			if (pack_mapped > peak_pack_mapped)
				peak_pack_mapped = pack_mapped;
			if (pack_open_windows > peak_pack_open_windows)
				peak_pack_open_windows = pack_open_windows;
774 775 776
			win->next = p->windows;
			p->windows = win;
		}
777
	}
778 779 780 781 782
	if (win != *w_cursor) {
		win->last_used = pack_used_ctr++;
		win->inuse_cnt++;
		*w_cursor = win;
	}
783
	offset -= win->offset;
784
	if (left)
785
		*left = win->len - xsize_t(offset);
786
	return win->base + offset;
787 788
}

789 790 791 792 793 794 795 796
static struct packed_git *alloc_packed_git(int extra)
{
	struct packed_git *p = xmalloc(sizeof(*p) + extra);
	memset(p, 0, sizeof(*p));
	p->pack_fd = -1;
	return p;
}

797
struct packed_git *add_packed_git(const char *path, int path_len, int local)
798 799
{
	struct stat st;
800
	struct packed_git *p = alloc_packed_git(path_len + 2);
801

802 803 804 805 806
	/*
	 * Make sure a corresponding .pack file exists and that
	 * the index looks sane.
	 */
	path_len -= strlen(".idx");
807 808
	if (path_len < 1) {
		free(p);
809
		return NULL;
810
	}
811
	memcpy(p->pack_name, path, path_len);
812 813 814 815 816

	strcpy(p->pack_name + path_len, ".keep");
	if (!access(p->pack_name, F_OK))
		p->pack_keep = 1;

817
	strcpy(p->pack_name + path_len, ".pack");
818
	if (stat(p->pack_name, &st) || !S_ISREG(st.st_mode)) {
819
		free(p);
820 821
		return NULL;
	}
822

823 824 825 826
	/* ok, it looks sane as far as we can check without
	 * actually mapping the pack file.
	 */
	p->pack_size = st.st_size;
827
	p->pack_local = local;
828
	p->mtime = st.st_mtime;
829 830
	if (path_len < 40 || get_sha1_hex(path + path_len - 40, p->sha1))
		hashclr(p->sha1);
831 832 833
	return p;
}

834
struct packed_git *parse_pack_index(unsigned char *sha1)
835
{
836
	const char *idx_path = sha1_pack_index_name(sha1);
837
	const char *path = sha1_pack_name(sha1);
838
	struct packed_git *p = alloc_packed_git(strlen(path) + 1);
839

840 841
	strcpy(p->pack_name, path);
	hashcpy(p->sha1, sha1);
842 843
	if (check_packed_git_idx(idx_path, p)) {
		free(p);
844
		return NULL;
845
	}
846 847 848 849 850 851 852 853 854 855

	return p;
}

void install_packed_git(struct packed_git *pack)
{
	pack->next = packed_git;
	packed_git = pack;
}

856
static void prepare_packed_git_one(char *objdir, int local)
857
{
858 859 860 861
	/* Ensure that this buffer is large enough so that we can
	   append "/pack/" without clobbering the stack even if
	   strlen(objdir) were PATH_MAX.  */
	char path[PATH_MAX + 1 + 4 + 1 + 1];
862 863 864 865 866 867 868
	int len;
	DIR *dir;
	struct dirent *de;

	sprintf(path, "%s/pack", objdir);
	len = strlen(path);
	dir = opendir(path);
869 870
	while (!dir && errno == EMFILE && unuse_one_window(packed_git, -1))
		dir = opendir(path);
871
	if (!dir) {
J
Junio C Hamano 已提交
872
		if (errno != ENOENT)
873
			error("unable to open object pack directory: %s: %s",
J
Junio C Hamano 已提交
874
			      path, strerror(errno));
875
		return;
876
	}
877 878 879 880 881
	path[len++] = '/';
	while ((de = readdir(dir)) != NULL) {
		int namelen = strlen(de->d_name);
		struct packed_git *p;

882
		if (!has_extension(de->d_name, ".idx"))
883 884
			continue;

885 886 887
		if (len + namelen + 1 > sizeof(path))
			continue;

888
		/* Don't reopen a pack we already have. */
889
		strcpy(path + len, de->d_name);
890 891 892 893 894 895
		for (p = packed_git; p; p = p->next) {
			if (!memcmp(path, p->pack_name, len + namelen - 4))
				break;
		}
		if (p)
			continue;
896 897 898
		/* See if it really is a valid .idx file with corresponding
		 * .pack file that we can map.
		 */
899
		p = add_packed_git(path, len + namelen, local);
900 901
		if (!p)
			continue;
902
		install_packed_git(p);
903
	}
904
	closedir(dir);
905 906
}

907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960
static int sort_pack(const void *a_, const void *b_)
{
	struct packed_git *a = *((struct packed_git **)a_);
	struct packed_git *b = *((struct packed_git **)b_);
	int st;

	/*
	 * Local packs tend to contain objects specific to our
	 * variant of the project than remote ones.  In addition,
	 * remote ones could be on a network mounted filesystem.
	 * Favor local ones for these reasons.
	 */
	st = a->pack_local - b->pack_local;
	if (st)
		return -st;

	/*
	 * Younger packs tend to contain more recent objects,
	 * and more recent objects tend to get accessed more
	 * often.
	 */
	if (a->mtime < b->mtime)
		return 1;
	else if (a->mtime == b->mtime)
		return 0;
	return -1;
}

static void rearrange_packed_git(void)
{
	struct packed_git **ary, *p;
	int i, n;

	for (n = 0, p = packed_git; p; p = p->next)
		n++;
	if (n < 2)
		return;

	/* prepare an array of packed_git for easier sorting */
	ary = xcalloc(n, sizeof(struct packed_git *));
	for (n = 0, p = packed_git; p; p = p->next)
		ary[n++] = p;

	qsort(ary, n, sizeof(struct packed_git *), sort_pack);

	/* link them back again */
	for (i = 0; i < n - 1; i++)
		ary[i]->next = ary[i + 1];
	ary[n - 1]->next = NULL;
	packed_git = ary[0];

	free(ary);
}

961
static int prepare_packed_git_run_once = 0;
962
void prepare_packed_git(void)
963
{
964
	struct alternate_object_database *alt;
965

966
	if (prepare_packed_git_run_once)
967
		return;
968
	prepare_packed_git_one(get_object_directory(), 1);
969
	prepare_alt_odb();
970
	for (alt = alt_odb_list; alt; alt = alt->next) {
971
		alt->name[-1] = 0;
972
		prepare_packed_git_one(alt->base, 0);
973
		alt->name[-1] = '/';
974
	}
975
	rearrange_packed_git();
976 977 978
	prepare_packed_git_run_once = 1;
}

979
void reprepare_packed_git(void)
980
{
981
	discard_revindex();
982 983
	prepare_packed_git_run_once = 0;
	prepare_packed_git();
984 985
}

986 987 988 989 990 991 992 993 994 995 996 997
static void mark_bad_packed_object(struct packed_git *p,
				   const unsigned char *sha1)
{
	unsigned i;
	for (i = 0; i < p->num_bad_objects; i++)
		if (!hashcmp(sha1, p->bad_object_sha1 + 20 * i))
			return;
	p->bad_object_sha1 = xrealloc(p->bad_object_sha1, 20 * (p->num_bad_objects + 1));
	hashcpy(p->bad_object_sha1 + 20 * p->num_bad_objects, sha1);
	p->num_bad_objects++;
}

998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009
static int has_packed_and_bad(const unsigned char *sha1)
{
	struct packed_git *p;
	unsigned i;

	for (p = packed_git; p; p = p->next)
		for (i = 0; i < p->num_bad_objects; i++)
			if (!hashcmp(sha1, p->bad_object_sha1 + 20 * i))
				return 1;
	return 0;
}

1010
int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
1011 1012
{
	unsigned char real_sha1[20];
1013
	hash_sha1_file(map, size, type, real_sha1);
1014
	return hashcmp(sha1, real_sha1) ? -1 : 0;
1015 1016
}

1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
static int git_open_noatime(const char *name)
{
	static int sha1_file_open_flag = O_NOATIME;
	int fd = open(name, O_RDONLY | sha1_file_open_flag);

	/* Might the failure be due to O_NOATIME? */
	if (fd < 0 && errno != ENOENT && sha1_file_open_flag) {
		fd = open(name, O_RDONLY);
		if (fd >= 0)
			sha1_file_open_flag = 0;
	}
	return fd;
}

static int open_sha1_file(const unsigned char *sha1)
{
	int fd;
	char *name = sha1_file_name(sha1);
	struct alternate_object_database *alt;

	fd = git_open_noatime(name);
	if (fd >= 0)
		return fd;

	prepare_alt_odb();
	errno = ENOENT;
	for (alt = alt_odb_list; alt; alt = alt->next) {
		name = alt->name;
		fill_sha1_path(name, sha1);
		fd = git_open_noatime(alt->base);
		if (fd >= 0)
			return fd;
	}
	return -1;
}

J
Junio C Hamano 已提交
1053
static void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
1054 1055
{
	void *map;
1056
	int fd;
1057

1058 1059 1060 1061
	fd = open_sha1_file(sha1);
	map = NULL;
	if (fd >= 0) {
		struct stat st;
1062

1063 1064 1065
		if (!fstat(fd, &st)) {
			*size = xsize_t(st.st_size);
			map = xmmap(NULL, *size, PROT_READ, MAP_PRIVATE, fd, 0);
1066
		}
1067
		close(fd);
1068 1069 1070 1071
	}
	return map;
}

1072
static int legacy_loose_object(unsigned char *map)
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
{
	unsigned int word;

	/*
	 * Is it a zlib-compressed buffer? If so, the first byte
	 * must be 0x78 (15-bit window size, deflated), and the
	 * first 16-bit word is evenly divisible by 31
	 */
	word = (map[0] << 8) + map[1];
	if (map[0] == 0x78 && !(word % 31))
		return 1;
	else
		return 0;
}

1088 1089
unsigned long unpack_object_header_buffer(const unsigned char *buf,
		unsigned long len, enum object_type *type, unsigned long *sizep)
1090
{
1091
	unsigned shift;
1092
	unsigned long size, c;
1093 1094 1095 1096 1097 1098 1099
	unsigned long used = 0;

	c = buf[used++];
	*type = (c >> 4) & 7;
	size = c & 15;
	shift = 4;
	while (c & 0x80) {
1100
		if (len <= used || bitsizeof(long) <= shift) {
1101
			error("bad object header");
1102
			return 0;
1103
		}
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
		c = buf[used++];
		size += (c & 0x7f) << shift;
		shift += 7;
	}
	*sizep = size;
	return used;
}

static int unpack_sha1_header(z_stream *stream, unsigned char *map, unsigned long mapsize, void *buffer, unsigned long bufsiz)
{
	unsigned long size, used;
	static const char valid_loose_object_type[8] = {
		0, /* OBJ_EXT */
		1, 1, 1, 1, /* "commit", "tree", "blob", "tag" */
		0, /* "delta" and others are invalid in a loose object */
1119
	};
1120
	enum object_type type;
1121

1122 1123 1124 1125 1126
	/* Get the data stream */
	memset(stream, 0, sizeof(*stream));
	stream->next_in = map;
	stream->avail_in = mapsize;
	stream->next_out = buffer;
1127 1128
	stream->avail_out = bufsiz;

1129
	if (legacy_loose_object(map)) {
1130 1131
		git_inflate_init(stream);
		return git_inflate(stream, 0);
1132 1133
	}

1134 1135 1136 1137 1138 1139 1140 1141

	/*
	 * There used to be a second loose object header format which
	 * was meant to mimic the in-pack format, allowing for direct
	 * copy of the object data.  This format turned up not to be
	 * really worth it and we don't write it any longer.  But we
	 * can still read it.
	 */
1142
	used = unpack_object_header_buffer(map, mapsize, &type, &size);
1143
	if (!used || !valid_loose_object_type[type])
1144
		return -1;
1145 1146
	map += used;
	mapsize -= used;
1147 1148 1149 1150

	/* Set up the stream for the rest.. */
	stream->next_in = map;
	stream->avail_in = mapsize;
1151
	git_inflate_init(stream);
1152 1153

	/* And generate the fake traditional header */
1154
	stream->total_out = 1 + snprintf(buffer, bufsiz, "%s %lu",
1155
					 typename(type), size);
1156
	return 0;
1157 1158
}

1159
static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size, const unsigned char *sha1)
1160 1161
{
	int bytes = strlen(buffer) + 1;
1162
	unsigned char *buf = xmallocz(size);
1163
	unsigned long n;
1164
	int status = Z_OK;
1165

1166 1167 1168 1169 1170
	n = stream->total_out - bytes;
	if (n > size)
		n = size;
	memcpy(buf, (char *) buffer + bytes, n);
	bytes = n;
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184
	if (bytes <= size) {
		/*
		 * The above condition must be (bytes <= size), not
		 * (bytes < size).  In other words, even though we
		 * expect no more output and set avail_out to zer0,
		 * the input zlib stream may have bytes that express
		 * "this concludes the stream", and we *do* want to
		 * eat that input.
		 *
		 * Otherwise we would not be able to test that we
		 * consumed all the input to reach the expected size;
		 * we also want to check that zlib tells us that all
		 * went well with status == Z_STREAM_END at the end.
		 */
1185 1186
		stream->next_out = buf + bytes;
		stream->avail_out = size - bytes;
1187
		while (status == Z_OK)
1188
			status = git_inflate(stream, Z_FINISH);
1189
	}
1190
	if (status == Z_STREAM_END && !stream->avail_in) {
1191
		git_inflate_end(stream);
1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
		return buf;
	}

	if (status < 0)
		error("corrupt loose object '%s'", sha1_to_hex(sha1));
	else if (stream->avail_in)
		error("garbage at end of loose object '%s'",
		      sha1_to_hex(sha1));
	free(buf);
	return NULL;
1202 1203 1204 1205 1206 1207 1208
}

/*
 * We used to just use "sscanf()", but that's actually way
 * too permissive for what we want to check. So do an anal
 * object header parse by hand.
 */
1209
static int parse_sha1_header(const char *hdr, unsigned long *sizep)
1210
{
1211
	char type[10];
1212 1213 1214 1215
	int i;
	unsigned long size;

	/*
J
Junio C Hamano 已提交
1216
	 * The type can be at most ten bytes (including the
1217
	 * terminating '\0' that we add), and is followed by
1218
	 * a space.
1219
	 */
1220
	i = 0;
1221 1222 1223 1224
	for (;;) {
		char c = *hdr++;
		if (c == ' ')
			break;
1225 1226
		type[i++] = c;
		if (i >= sizeof(type))
1227 1228
			return -1;
	}
1229
	type[i] = 0;
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251

	/*
	 * The length must follow immediately, and be in canonical
	 * decimal format (ie "010" is not valid).
	 */
	size = *hdr++ - '0';
	if (size > 9)
		return -1;
	if (size) {
		for (;;) {
			unsigned long c = *hdr - '0';
			if (c > 9)
				break;
			hdr++;
			size = size * 10 + c;
		}
	}
	*sizep = size;

	/*
	 * The length must be followed by a zero byte
	 */
1252
	return *hdr ? -1 : type_from_string(type);
1253 1254
}

1255
static void *unpack_sha1_file(void *map, unsigned long mapsize, enum object_type *type, unsigned long *size, const unsigned char *sha1)
1256
{
1257
	int ret;
1258
	z_stream stream;
1259
	char hdr[8192];
1260

1261
	ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
1262
	if (ret < Z_OK || (*type = parse_sha1_header(hdr, size)) < 0)
1263 1264
		return NULL;

1265
	return unpack_sha1_rest(&stream, hdr, *size, sha1);
1266 1267
}

N
Nicolas Pitre 已提交
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
unsigned long get_size_from_delta(struct packed_git *p,
				  struct pack_window **w_curs,
			          off_t curpos)
{
	const unsigned char *data;
	unsigned char delta_head[20], *in;
	z_stream stream;
	int st;

	memset(&stream, 0, sizeof(stream));
	stream.next_out = delta_head;
	stream.avail_out = sizeof(delta_head);

1281
	git_inflate_init(&stream);
N
Nicolas Pitre 已提交
1282 1283 1284
	do {
		in = use_pack(p, w_curs, curpos, &stream.avail_in);
		stream.next_in = in;
1285
		st = git_inflate(&stream, Z_FINISH);
N
Nicolas Pitre 已提交
1286 1287 1288
		curpos += stream.next_in - in;
	} while ((st == Z_OK || st == Z_BUF_ERROR) &&
		 stream.total_out < sizeof(delta_head));
1289
	git_inflate_end(&stream);
1290 1291 1292 1293
	if ((st != Z_STREAM_END) && stream.total_out != sizeof(delta_head)) {
		error("delta data unpack-initial failed");
		return 0;
	}
N
Nicolas Pitre 已提交
1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306

	/* Examine the initial part of the delta to figure out
	 * the result size.
	 */
	data = delta_head;

	/* ignore base size */
	get_delta_hdr_size(&data, delta_head+sizeof(delta_head));

	/* Read the result size */
	return get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
}

1307
static off_t get_delta_base(struct packed_git *p,
1308
				    struct pack_window **w_curs,
1309
				    off_t *curpos,
1310
				    enum object_type type,
1311
				    off_t delta_obj_offset)
1312
{
1313
	unsigned char *base_info = use_pack(p, w_curs, *curpos, NULL);
1314
	off_t base_offset;
1315

1316 1317 1318 1319 1320 1321
	/* use_pack() assured us we have [base_info, base_info + 20)
	 * as a range that we can look at without walking off the
	 * end of the mapped window.  Its actually the hash size
	 * that is assured.  An OFS_DELTA longer than the hash size
	 * is stupid, as then a REF_DELTA would be smaller to store.
	 */
1322
	if (type == OBJ_OFS_DELTA) {
1323 1324 1325 1326 1327
		unsigned used = 0;
		unsigned char c = base_info[used++];
		base_offset = c & 127;
		while (c & 128) {
			base_offset += 1;
1328
			if (!base_offset || MSB(base_offset, 7))
1329
				return 0;  /* overflow */
1330 1331 1332 1333
			c = base_info[used++];
			base_offset = (base_offset << 7) + (c & 127);
		}
		base_offset = delta_obj_offset - base_offset;
1334
		if (base_offset <= 0 || base_offset >= delta_obj_offset)
1335
			return 0;  /* out of bound */
1336
		*curpos += used;
1337
	} else if (type == OBJ_REF_DELTA) {
1338 1339
		/* The base entry _must_ be in the same pack */
		base_offset = find_pack_entry_one(base_info, p);
1340
		*curpos += 20;
1341 1342
	} else
		die("I am totally screwed");
1343
	return base_offset;
1344 1345
}

J
Junio C Hamano 已提交
1346
/* forward declaration for a mutually recursive function */
1347
static int packed_object_info(struct packed_git *p, off_t offset,
1348
			      unsigned long *sizep);
J
Junio C Hamano 已提交
1349

N
Nicolas Pitre 已提交
1350
static int packed_delta_info(struct packed_git *p,
1351
			     struct pack_window **w_curs,
1352
			     off_t curpos,
1353
			     enum object_type type,
1354
			     off_t obj_offset,
N
Nicolas Pitre 已提交
1355
			     unsigned long *sizep)
1356
{
1357
	off_t base_offset;
J
Junio C Hamano 已提交
1358

1359
	base_offset = get_delta_base(p, w_curs, &curpos, type, obj_offset);
1360 1361
	if (!base_offset)
		return OBJ_BAD;
1362
	type = packed_object_info(p, base_offset, NULL);
1363
	if (type <= OBJ_NONE) {
1364 1365 1366 1367 1368 1369
		struct revindex_entry *revidx;
		const unsigned char *base_sha1;
		revidx = find_pack_revindex(p, base_offset);
		if (!revidx)
			return OBJ_BAD;
		base_sha1 = nth_packed_object_sha1(p, revidx->nr);
1370 1371 1372 1373 1374
		mark_bad_packed_object(p, base_sha1);
		type = sha1_object_info(base_sha1, NULL);
		if (type <= OBJ_NONE)
			return OBJ_BAD;
	}
J
Junio C Hamano 已提交
1375

1376 1377 1378 1379 1380
	/* We choose to only get the type of the base object and
	 * ignore potentially corrupt pack file that expects the delta
	 * based on a base with a wrong size.  This saves tons of
	 * inflate() calls.
	 */
1381
	if (sizep) {
N
Nicolas Pitre 已提交
1382
		*sizep = get_size_from_delta(p, w_curs, curpos);
1383 1384 1385
		if (*sizep == 0)
			type = OBJ_BAD;
	}
1386 1387

	return type;
1388 1389
}

1390 1391
static int unpack_object_header(struct packed_git *p,
				struct pack_window **w_curs,
1392
				off_t *curpos,
1393
				unsigned long *sizep)
1394
{
1395 1396
	unsigned char *base;
	unsigned int left;
1397
	unsigned long used;
1398
	enum object_type type;
1399

1400 1401
	/* use_pack() assures us we have [base, base + 20) available
	 * as a range that we can look at at.  (Its actually the hash
P
Pavel Roskin 已提交
1402
	 * size that is assured.)  With our object header encoding
1403 1404 1405
	 * the maximum deflated object size is 2^137, which is just
	 * insane, so we know won't exceed what we have been given.
	 */
1406
	base = use_pack(p, w_curs, *curpos, &left);
1407 1408 1409 1410 1411
	used = unpack_object_header_buffer(base, left, &type, sizep);
	if (!used) {
		type = OBJ_BAD;
	} else
		*curpos += used;
1412

1413
	return type;
1414 1415
}

1416
const char *packed_object_info_detail(struct packed_git *p,
1417
				      off_t obj_offset,
1418 1419 1420 1421
				      unsigned long *size,
				      unsigned long *store_size,
				      unsigned int *delta_chain_length,
				      unsigned char *base_sha1)
1422
{
1423
	struct pack_window *w_curs = NULL;
1424 1425
	off_t curpos;
	unsigned long dummy;
N
Nicolas Pitre 已提交
1426
	unsigned char *next_sha1;
1427
	enum object_type type;
1428
	struct revindex_entry *revidx;
1429

N
Nicolas Pitre 已提交
1430
	*delta_chain_length = 0;
1431
	curpos = obj_offset;
1432
	type = unpack_object_header(p, &w_curs, &curpos, size);
N
Nicolas Pitre 已提交
1433

1434 1435 1436
	revidx = find_pack_revindex(p, obj_offset);
	*store_size = revidx[1].offset - obj_offset;

N
Nicolas Pitre 已提交
1437
	for (;;) {
1438
		switch (type) {
N
Nicolas Pitre 已提交
1439
		default:
1440
			die("pack %s contains unknown object type %d",
1441
			    p->pack_name, type);
N
Nicolas Pitre 已提交
1442 1443 1444 1445
		case OBJ_COMMIT:
		case OBJ_TREE:
		case OBJ_BLOB:
		case OBJ_TAG:
1446
			unuse_pack(&w_curs);
1447
			return typename(type);
1448
		case OBJ_OFS_DELTA:
1449
			obj_offset = get_delta_base(p, &w_curs, &curpos, type, obj_offset);
1450 1451 1452
			if (!obj_offset)
				die("pack %s contains bad delta base reference of type %s",
				    p->pack_name, typename(type));
1453
			if (*delta_chain_length == 0) {
1454 1455
				revidx = find_pack_revindex(p, obj_offset);
				hashcpy(base_sha1, nth_packed_object_sha1(p, revidx->nr));
1456 1457 1458
			}
			break;
		case OBJ_REF_DELTA:
1459
			next_sha1 = use_pack(p, &w_curs, curpos, NULL);
N
Nicolas Pitre 已提交
1460 1461
			if (*delta_chain_length == 0)
				hashcpy(base_sha1, next_sha1);
1462
			obj_offset = find_pack_entry_one(next_sha1, p);
N
Nicolas Pitre 已提交
1463 1464 1465
			break;
		}
		(*delta_chain_length)++;
1466
		curpos = obj_offset;
1467
		type = unpack_object_header(p, &w_curs, &curpos, &dummy);
1468 1469 1470
	}
}

1471
static int packed_object_info(struct packed_git *p, off_t obj_offset,
1472
			      unsigned long *sizep)
1473
{
1474
	struct pack_window *w_curs = NULL;
1475 1476
	unsigned long size;
	off_t curpos = obj_offset;
1477
	enum object_type type;
1478

1479
	type = unpack_object_header(p, &w_curs, &curpos, &size);
1480

1481
	switch (type) {
1482 1483
	case OBJ_OFS_DELTA:
	case OBJ_REF_DELTA:
1484 1485
		type = packed_delta_info(p, &w_curs, curpos,
					 type, obj_offset, sizep);
1486
		break;
1487 1488 1489 1490
	case OBJ_COMMIT:
	case OBJ_TREE:
	case OBJ_BLOB:
	case OBJ_TAG:
1491 1492
		if (sizep)
			*sizep = size;
1493
		break;
1494
	default:
1495 1496 1497
		error("unknown object type %i at offset %"PRIuMAX" in %s",
		      type, (uintmax_t)obj_offset, p->pack_name);
		type = OBJ_BAD;
1498
	}
1499
	unuse_pack(&w_curs);
1500
	return type;
1501 1502
}

1503
static void *unpack_compressed_entry(struct packed_git *p,
1504
				    struct pack_window **w_curs,
1505
				    off_t curpos,
1506
				    unsigned long size)
1507 1508 1509
{
	int st;
	z_stream stream;
1510
	unsigned char *buffer, *in;
1511

1512
	buffer = xmallocz(size);
1513 1514
	memset(&stream, 0, sizeof(stream));
	stream.next_out = buffer;
1515
	stream.avail_out = size + 1;
1516

1517
	git_inflate_init(&stream);
1518
	do {
1519
		in = use_pack(p, w_curs, curpos, &stream.avail_in);
1520
		stream.next_in = in;
1521
		st = git_inflate(&stream, Z_FINISH);
1522 1523
		if (!stream.avail_out)
			break; /* the payload is larger than it should be */
1524
		curpos += stream.next_in - in;
1525
	} while (st == Z_OK || st == Z_BUF_ERROR);
1526
	git_inflate_end(&stream);
1527 1528 1529 1530 1531 1532 1533 1534
	if ((st != Z_STREAM_END) || stream.total_out != size) {
		free(buffer);
		return NULL;
	}

	return buffer;
}

1535 1536
#define MAX_DELTA_CACHE (256)

1537
static size_t delta_base_cached;
1538 1539 1540 1541 1542 1543

static struct delta_base_cache_lru_list {
	struct delta_base_cache_lru_list *prev;
	struct delta_base_cache_lru_list *next;
} delta_base_cache_lru = { &delta_base_cache_lru, &delta_base_cache_lru };

1544
static struct delta_base_cache_entry {
1545 1546
	struct delta_base_cache_lru_list lru;
	void *data;
1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558
	struct packed_git *p;
	off_t base_offset;
	unsigned long size;
	enum object_type type;
} delta_base_cache[MAX_DELTA_CACHE];

static unsigned long pack_entry_hash(struct packed_git *p, off_t base_offset)
{
	unsigned long hash;

	hash = (unsigned long)p + (unsigned long)base_offset;
	hash += (hash >> 8) + (hash >> 16);
1559
	return hash % MAX_DELTA_CACHE;
1560 1561
}

1562
static void *cache_or_unpack_entry(struct packed_git *p, off_t base_offset,
1563
	unsigned long *base_size, enum object_type *type, int keep_cache)
1564
{
1565 1566 1567 1568 1569
	void *ret;
	unsigned long hash = pack_entry_hash(p, base_offset);
	struct delta_base_cache_entry *ent = delta_base_cache + hash;

	ret = ent->data;
1570 1571
	if (!ret || ent->p != p || ent->base_offset != base_offset)
		return unpack_entry(p, base_offset, type, base_size);
1572

1573
	if (!keep_cache) {
1574
		ent->data = NULL;
1575 1576
		ent->lru.next->prev = ent->lru.prev;
		ent->lru.prev->next = ent->lru.next;
1577
		delta_base_cached -= ent->size;
P
Pierre Habouzit 已提交
1578 1579
	} else {
		ret = xmemdupz(ent->data, ent->size);
1580
	}
1581 1582 1583
	*type = ent->type;
	*base_size = ent->size;
	return ret;
1584 1585
}

1586 1587 1588 1589 1590
static inline void release_delta_base_cache(struct delta_base_cache_entry *ent)
{
	if (ent->data) {
		free(ent->data);
		ent->data = NULL;
1591 1592
		ent->lru.next->prev = ent->lru.prev;
		ent->lru.prev->next = ent->lru.next;
1593 1594 1595 1596
		delta_base_cached -= ent->size;
	}
}

1597 1598 1599 1600 1601 1602 1603
void clear_delta_base_cache(void)
{
	unsigned long p;
	for (p = 0; p < MAX_DELTA_CACHE; p++)
		release_delta_base_cache(&delta_base_cache[p]);
}

1604 1605 1606
static void add_delta_base_cache(struct packed_git *p, off_t base_offset,
	void *base, unsigned long base_size, enum object_type type)
{
1607
	unsigned long hash = pack_entry_hash(p, base_offset);
1608
	struct delta_base_cache_entry *ent = delta_base_cache + hash;
1609
	struct delta_base_cache_lru_list *lru;
1610

1611 1612
	release_delta_base_cache(ent);
	delta_base_cached += base_size;
1613 1614 1615 1616 1617 1618

	for (lru = delta_base_cache_lru.next;
	     delta_base_cached > delta_base_cache_limit
	     && lru != &delta_base_cache_lru;
	     lru = lru->next) {
		struct delta_base_cache_entry *f = (void *)lru;
1619 1620 1621
		if (f->type == OBJ_BLOB)
			release_delta_base_cache(f);
	}
1622 1623 1624 1625 1626 1627 1628
	for (lru = delta_base_cache_lru.next;
	     delta_base_cached > delta_base_cache_limit
	     && lru != &delta_base_cache_lru;
	     lru = lru->next) {
		struct delta_base_cache_entry *f = (void *)lru;
		release_delta_base_cache(f);
	}
1629

1630 1631 1632 1633 1634
	ent->p = p;
	ent->base_offset = base_offset;
	ent->type = type;
	ent->data = base;
	ent->size = base_size;
1635 1636 1637 1638
	ent->lru.next = &delta_base_cache_lru;
	ent->lru.prev = delta_base_cache_lru.prev;
	delta_base_cache_lru.prev->next = &ent->lru;
	delta_base_cache_lru.prev = &ent->lru;
1639 1640
}

1641 1642 1643
static void *read_object(const unsigned char *sha1, enum object_type *type,
			 unsigned long *size);

1644
static void *unpack_delta_entry(struct packed_git *p,
1645
				struct pack_window **w_curs,
1646
				off_t curpos,
1647
				unsigned long delta_size,
1648
				off_t obj_offset,
1649
				enum object_type *type,
1650
				unsigned long *sizep)
1651
{
1652
	void *delta_data, *result, *base;
1653 1654
	unsigned long base_size;
	off_t base_offset;
N
Nicolas Pitre 已提交
1655

1656
	base_offset = get_delta_base(p, w_curs, &curpos, *type, obj_offset);
1657 1658 1659 1660 1661 1662
	if (!base_offset) {
		error("failed to validate delta base reference "
		      "at offset %"PRIuMAX" from %s",
		      (uintmax_t)curpos, p->pack_name);
		return NULL;
	}
1663
	unuse_pack(w_curs);
1664
	base = cache_or_unpack_entry(p, base_offset, &base_size, type, 0);
1665 1666 1667 1668 1669 1670 1671
	if (!base) {
		/*
		 * We're probably in deep shit, but let's try to fetch
		 * the required base anyway from another pack or loose.
		 * This is costly but should happen only in the presence
		 * of a corrupted pack, and is better than failing outright.
		 */
1672 1673 1674 1675 1676 1677
		struct revindex_entry *revidx;
		const unsigned char *base_sha1;
		revidx = find_pack_revindex(p, base_offset);
		if (!revidx)
			return NULL;
		base_sha1 = nth_packed_object_sha1(p, revidx->nr);
1678 1679 1680 1681 1682
		error("failed to read delta base object %s"
		      " at offset %"PRIuMAX" from %s",
		      sha1_to_hex(base_sha1), (uintmax_t)base_offset,
		      p->pack_name);
		mark_bad_packed_object(p, base_sha1);
1683
		base = read_object(base_sha1, type, &base_size);
1684 1685 1686
		if (!base)
			return NULL;
	}
1687

1688
	delta_data = unpack_compressed_entry(p, w_curs, curpos, delta_size);
1689 1690 1691 1692 1693 1694 1695
	if (!delta_data) {
		error("failed to unpack compressed delta "
		      "at offset %"PRIuMAX" from %s",
		      (uintmax_t)curpos, p->pack_name);
		free(base);
		return NULL;
	}
1696 1697
	result = patch_delta(base, base_size,
			     delta_data, delta_size,
1698
			     sizep);
1699 1700 1701
	if (!result)
		die("failed to apply delta");
	free(delta_data);
1702
	add_delta_base_cache(p, base_offset, base, base_size, *type);
1703 1704 1705
	return result;
}

1706 1707
int do_check_packed_object_crc;

1708
void *unpack_entry(struct packed_git *p, off_t obj_offset,
1709
		   enum object_type *type, unsigned long *sizep)
1710
{
1711
	struct pack_window *w_curs = NULL;
1712
	off_t curpos = obj_offset;
1713
	void *data;
1714

1715 1716 1717 1718 1719 1720 1721 1722 1723
	if (do_check_packed_object_crc && p->index_version > 1) {
		struct revindex_entry *revidx = find_pack_revindex(p, obj_offset);
		unsigned long len = revidx[1].offset - obj_offset;
		if (check_pack_crc(p, &w_curs, obj_offset, len, revidx->nr)) {
			const unsigned char *sha1 =
				nth_packed_object_sha1(p, revidx->nr);
			error("bad packed object CRC for %s",
			      sha1_to_hex(sha1));
			mark_bad_packed_object(p, sha1);
1724
			unuse_pack(&w_curs);
1725 1726 1727 1728
			return NULL;
		}
	}

1729 1730
	*type = unpack_object_header(p, &w_curs, &curpos, sizep);
	switch (*type) {
1731 1732
	case OBJ_OFS_DELTA:
	case OBJ_REF_DELTA:
1733 1734
		data = unpack_delta_entry(p, &w_curs, curpos, *sizep,
					  obj_offset, type, sizep);
1735
		break;
1736 1737 1738 1739
	case OBJ_COMMIT:
	case OBJ_TREE:
	case OBJ_BLOB:
	case OBJ_TAG:
1740
		data = unpack_compressed_entry(p, &w_curs, curpos, *sizep);
1741
		break;
1742
	default:
1743 1744 1745
		data = NULL;
		error("unknown object type %i at offset %"PRIuMAX" in %s",
		      *type, (uintmax_t)obj_offset, p->pack_name);
1746
	}
1747
	unuse_pack(&w_curs);
1748
	return data;
1749 1750
}

1751
const unsigned char *nth_packed_object_sha1(struct packed_git *p,
1752
					    uint32_t n)
1753
{
1754
	const unsigned char *index = p->index_data;
1755 1756 1757 1758 1759
	if (!index) {
		if (open_pack_index(p))
			return NULL;
		index = p->index_data;
	}
N
Nicolas Pitre 已提交
1760
	if (n >= p->num_objects)
1761
		return NULL;
1762 1763 1764 1765 1766 1767 1768 1769 1770
	index += 4 * 256;
	if (p->index_version == 1) {
		return index + 24 * n + 4;
	} else {
		index += 8;
		return index + 20 * n;
	}
}

N
Nicolas Pitre 已提交
1771
off_t nth_packed_object_offset(const struct packed_git *p, uint32_t n)
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786
{
	const unsigned char *index = p->index_data;
	index += 4 * 256;
	if (p->index_version == 1) {
		return ntohl(*((uint32_t *)(index + 24 * n)));
	} else {
		uint32_t off;
		index += 8 + p->num_objects * (20 + 4);
		off = ntohl(*((uint32_t *)(index + 4 * n)));
		if (!(off & 0x80000000))
			return off;
		index += p->num_objects * 4 + (off & 0x7fffffff) * 8;
		return (((uint64_t)ntohl(*((uint32_t *)(index + 0)))) << 32) |
				   ntohl(*((uint32_t *)(index + 4)));
	}
1787 1788
}

1789
off_t find_pack_entry_one(const unsigned char *sha1,
N
Nicolas Pitre 已提交
1790
				  struct packed_git *p)
1791
{
1792 1793
	const uint32_t *level1_ofs = p->index_data;
	const unsigned char *index = p->index_data;
1794 1795 1796 1797 1798 1799
	unsigned hi, lo, stride;
	static int use_lookup = -1;
	static int debug_lookup = -1;

	if (debug_lookup < 0)
		debug_lookup = !!getenv("GIT_DEBUG_LOOKUP");
1800

1801 1802 1803 1804 1805 1806
	if (!index) {
		if (open_pack_index(p))
			return 0;
		level1_ofs = p->index_data;
		index = p->index_data;
	}
1807 1808 1809 1810
	if (p->index_version > 1) {
		level1_ofs += 2;
		index += 8;
	}
1811
	index += 4 * 256;
1812 1813
	hi = ntohl(level1_ofs[*sha1]);
	lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
1814 1815 1816 1817 1818 1819 1820 1821
	if (p->index_version > 1) {
		stride = 20;
	} else {
		stride = 24;
		index += 4;
	}

	if (debug_lookup)
1822
		printf("%02x%02x%02x... lo %u hi %u nr %"PRIu32"\n",
1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833
		       sha1[0], sha1[1], sha1[2], lo, hi, p->num_objects);

	if (use_lookup < 0)
		use_lookup = !!getenv("GIT_USE_LOOKUP");
	if (use_lookup) {
		int pos = sha1_entry_pos(index, stride, 0,
					 lo, hi, p->num_objects, sha1);
		if (pos < 0)
			return 0;
		return nth_packed_object_offset(p, pos);
	}
1834 1835

	do {
1836
		unsigned mi = (lo + hi) / 2;
1837 1838 1839 1840 1841
		int cmp = hashcmp(index + mi * stride, sha1);

		if (debug_lookup)
			printf("lo %u hi %u rg %u mi %u\n",
			       lo, hi, hi - lo, mi);
N
Nicolas Pitre 已提交
1842
		if (!cmp)
1843
			return nth_packed_object_offset(p, mi);
1844 1845 1846 1847 1848 1849 1850 1851
		if (cmp > 0)
			hi = mi;
		else
			lo = mi+1;
	} while (lo < hi);
	return 0;
}

1852
static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
1853
{
1854
	static struct packed_git *last_found = (void *)1;
1855
	struct packed_git *p;
1856
	off_t offset;
N
Nicolas Pitre 已提交
1857

1858
	prepare_packed_git();
1859 1860 1861
	if (!packed_git)
		return 0;
	p = (last_found == (void *)1) ? packed_git : last_found;
1862

1863
	do {
1864 1865 1866 1867 1868 1869 1870
		if (p->num_bad_objects) {
			unsigned i;
			for (i = 0; i < p->num_bad_objects; i++)
				if (!hashcmp(sha1, p->bad_object_sha1 + 20 * i))
					goto next;
		}

N
Nicolas Pitre 已提交
1871 1872
		offset = find_pack_entry_one(sha1, p);
		if (offset) {
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882
			/*
			 * We are about to tell the caller where they can
			 * locate the requested object.  We better make
			 * sure the packfile is still here and can be
			 * accessed before supplying that answer, as
			 * it may have been deleted since the index
			 * was loaded!
			 */
			if (p->pack_fd == -1 && open_packed_git(p)) {
				error("packfile %s cannot be accessed", p->pack_name);
1883
				goto next;
1884
			}
N
Nicolas Pitre 已提交
1885 1886 1887
			e->offset = offset;
			e->p = p;
			hashcpy(e->sha1, sha1);
1888
			last_found = p;
1889
			return 1;
N
Nicolas Pitre 已提交
1890
		}
1891 1892 1893 1894 1895 1896 1897 1898 1899

		next:
		if (p == last_found)
			p = packed_git;
		else
			p = p->next;
		if (p == last_found)
			p = p->next;
	} while (p);
1900 1901 1902
	return 0;
}

J
Junio C Hamano 已提交
1903
struct packed_git *find_sha1_pack(const unsigned char *sha1,
1904 1905 1906 1907 1908
				  struct packed_git *packs)
{
	struct packed_git *p;

	for (p = packs; p; p = p->next) {
N
Nicolas Pitre 已提交
1909
		if (find_pack_entry_one(sha1, p))
1910 1911 1912
			return p;
	}
	return NULL;
1913

1914 1915
}

1916
static int sha1_loose_object_info(const unsigned char *sha1, unsigned long *sizep)
1917
{
1918
	int status;
1919 1920 1921
	unsigned long mapsize, size;
	void *map;
	z_stream stream;
N
Nicolas Pitre 已提交
1922
	char hdr[32];
1923

1924
	map = map_sha1_file(sha1, &mapsize);
1925 1926
	if (!map)
		return error("unable to find %s", sha1_to_hex(sha1));
1927 1928 1929
	if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
		status = error("unable to unpack %s header",
			       sha1_to_hex(sha1));
1930
	else if ((status = parse_sha1_header(hdr, &size)) < 0)
1931
		status = error("unable to parse %s header", sha1_to_hex(sha1));
1932 1933
	else if (sizep)
		*sizep = size;
1934
	git_inflate_end(&stream);
1935 1936 1937 1938
	munmap(map, mapsize);
	return status;
}

1939
int sha1_object_info(const unsigned char *sha1, unsigned long *sizep)
1940 1941
{
	struct pack_entry e;
1942
	int status;
1943

1944
	if (!find_pack_entry(sha1, &e)) {
1945 1946 1947 1948 1949 1950
		/* Most likely it's a loose object. */
		status = sha1_loose_object_info(sha1, sizep);
		if (status >= 0)
			return status;

		/* Not a loose object; someone else may have just packed it. */
1951
		reprepare_packed_git();
1952
		if (!find_pack_entry(sha1, &e))
1953
			return status;
1954
	}
1955 1956 1957 1958 1959 1960 1961 1962

	status = packed_object_info(e.p, e.offset, sizep);
	if (status < 0) {
		mark_bad_packed_object(e.p, sha1);
		status = sha1_object_info(sha1, sizep);
	}

	return status;
1963 1964
}

1965 1966
static void *read_packed_sha1(const unsigned char *sha1,
			      enum object_type *type, unsigned long *size)
1967 1968
{
	struct pack_entry e;
1969
	void *data;
1970

1971
	if (!find_pack_entry(sha1, &e))
1972
		return NULL;
1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983
	data = cache_or_unpack_entry(e.p, e.offset, size, type, 1);
	if (!data) {
		/*
		 * We're probably in deep shit, but let's try to fetch
		 * the required object anyway from another pack or loose.
		 * This should happen only in the presence of a corrupted
		 * pack, and is better than failing outright.
		 */
		error("failed to read object %s at offset %"PRIuMAX" from %s",
		      sha1_to_hex(sha1), (uintmax_t)e.offset, e.p->pack_name);
		mark_bad_packed_object(e.p, sha1);
1984
		data = read_object(sha1, type, size);
1985 1986
	}
	return data;
1987 1988
}

1989 1990 1991 1992 1993 1994 1995 1996
/*
 * This is meant to hold a *small* number of objects that you would
 * want read_sha1_file() to be able to return, but yet you do not want
 * to write them into the object store (e.g. a browse-only
 * application).
 */
static struct cached_object {
	unsigned char sha1[20];
1997
	enum object_type type;
1998 1999 2000 2001 2002
	void *buf;
	unsigned long size;
} *cached_objects;
static int cached_object_nr, cached_object_alloc;

J
Jeff King 已提交
2003
static struct cached_object empty_tree = {
J
Jeff King 已提交
2004
	EMPTY_TREE_SHA1_BIN,
J
Jeff King 已提交
2005 2006 2007 2008 2009
	OBJ_TREE,
	"",
	0
};

2010 2011 2012 2013 2014 2015 2016 2017 2018
static struct cached_object *find_cached_object(const unsigned char *sha1)
{
	int i;
	struct cached_object *co = cached_objects;

	for (i = 0; i < cached_object_nr; i++, co++) {
		if (!hashcmp(co->sha1, sha1))
			return co;
	}
J
Jeff King 已提交
2019 2020
	if (!hashcmp(sha1, empty_tree.sha1))
		return &empty_tree;
2021 2022 2023
	return NULL;
}

2024 2025
int pretend_sha1_file(void *buf, unsigned long len, enum object_type type,
		      unsigned char *sha1)
2026 2027 2028
{
	struct cached_object *co;

2029
	hash_sha1_file(buf, len, typename(type), sha1);
2030 2031 2032 2033 2034 2035 2036 2037 2038 2039
	if (has_sha1_file(sha1) || find_cached_object(sha1))
		return 0;
	if (cached_object_alloc <= cached_object_nr) {
		cached_object_alloc = alloc_nr(cached_object_alloc);
		cached_objects = xrealloc(cached_objects,
					  sizeof(*cached_objects) *
					  cached_object_alloc);
	}
	co = &cached_objects[cached_object_nr++];
	co->size = len;
2040
	co->type = type;
J
Junio C Hamano 已提交
2041 2042
	co->buf = xmalloc(len);
	memcpy(co->buf, buf, len);
2043 2044 2045 2046
	hashcpy(co->sha1, sha1);
	return 0;
}

2047 2048
static void *read_object(const unsigned char *sha1, enum object_type *type,
			 unsigned long *size)
2049 2050 2051
{
	unsigned long mapsize;
	void *map, *buf;
2052 2053 2054 2055
	struct cached_object *co;

	co = find_cached_object(sha1);
	if (co) {
2056
		*type = co->type;
2057
		*size = co->size;
P
Pierre Habouzit 已提交
2058
		return xmemdupz(co->buf, co->size);
2059
	}
2060

2061 2062 2063
	buf = read_packed_sha1(sha1, type, size);
	if (buf)
		return buf;
2064
	map = map_sha1_file(sha1, &mapsize);
2065
	if (map) {
2066
		buf = unpack_sha1_file(map, mapsize, type, size, sha1);
2067 2068 2069
		munmap(map, mapsize);
		return buf;
	}
2070
	reprepare_packed_git();
2071
	return read_packed_sha1(sha1, type, size);
2072 2073
}

2074 2075 2076 2077
void *read_sha1_file_repl(const unsigned char *sha1,
			  enum object_type *type,
			  unsigned long *size,
			  const unsigned char **replacement)
2078
{
2079 2080 2081 2082 2083 2084 2085 2086
	const unsigned char *repl = lookup_replace_object(sha1);
	void *data = read_object(repl, type, size);

	/* die if we replaced an object with one that does not exist */
	if (!data && repl != sha1)
		die("replacement %s not found for %s",
		    sha1_to_hex(repl), sha1_to_hex(sha1));

2087
	/* legacy behavior is to die on corrupted objects */
2088 2089 2090
	if (!data && (has_loose_object(repl) || has_packed_and_bad(repl)))
		die("object %s is corrupted", sha1_to_hex(repl));

2091 2092 2093
	if (replacement)
		*replacement = repl;

2094 2095 2096
	return data;
}

2097
void *read_object_with_reference(const unsigned char *sha1,
2098
				 const char *required_type_name,
2099 2100
				 unsigned long *size,
				 unsigned char *actual_sha1_return)
2101
{
2102
	enum object_type type, required_type;
2103 2104
	void *buffer;
	unsigned long isize;
2105
	unsigned char actual_sha1[20];
2106

2107
	required_type = type_from_string(required_type_name);
2108
	hashcpy(actual_sha1, sha1);
2109 2110 2111
	while (1) {
		int ref_length = -1;
		const char *ref_type = NULL;
2112

2113
		buffer = read_sha1_file(actual_sha1, &type, &isize);
2114 2115
		if (!buffer)
			return NULL;
2116
		if (type == required_type) {
2117 2118
			*size = isize;
			if (actual_sha1_return)
2119
				hashcpy(actual_sha1_return, actual_sha1);
2120 2121 2122
			return buffer;
		}
		/* Handle references */
2123
		else if (type == OBJ_COMMIT)
2124
			ref_type = "tree ";
2125
		else if (type == OBJ_TAG)
2126 2127 2128 2129 2130 2131
			ref_type = "object ";
		else {
			free(buffer);
			return NULL;
		}
		ref_length = strlen(ref_type);
2132

2133 2134
		if (ref_length + 40 > isize ||
		    memcmp(buffer, ref_type, ref_length) ||
2135
		    get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
2136 2137 2138
			free(buffer);
			return NULL;
		}
2139
		free(buffer);
2140 2141
		/* Now we have the ID of the referred-to object in
		 * actual_sha1.  Check again. */
2142 2143 2144
	}
}

N
Nicolas Pitre 已提交
2145
static void write_sha1_file_prepare(const void *buf, unsigned long len,
R
Rene Scharfe 已提交
2146
                                    const char *type, unsigned char *sha1,
N
Nicolas Pitre 已提交
2147
                                    char *hdr, int *hdrlen)
2148
{
2149
	git_SHA_CTX c;
2150 2151

	/* Generate the header */
N
Nicolas Pitre 已提交
2152
	*hdrlen = sprintf(hdr, "%s %lu", type, len)+1;
2153 2154

	/* Sha1.. */
2155 2156 2157 2158
	git_SHA1_Init(&c);
	git_SHA1_Update(&c, hdr, *hdrlen);
	git_SHA1_Update(&c, buf, len);
	git_SHA1_Final(sha1, &c);
2159 2160
}

2161
/*
2162 2163 2164 2165
 * Move the just written object into its final resting place.
 * NEEDSWORK: this should be renamed to finalize_temp_file() as
 * "moving" is only a part of what it does, when no patch between
 * master to pu changes the call sites of this function.
2166
 */
2167
int move_temp_to_file(const char *tmpfile, const char *filename)
2168
{
2169
	int ret = 0;
2170

2171
	if (object_creation_mode == OBJECT_CREATION_USES_RENAMES)
2172 2173
		goto try_rename;
	else if (link(tmpfile, filename))
2174
		ret = errno;
2175 2176 2177 2178 2179 2180 2181 2182 2183

	/*
	 * Coda hack - coda doesn't like cross-directory links,
	 * so we fall back to a rename, which will mean that it
	 * won't be able to check collisions, but that's not a
	 * big deal.
	 *
	 * The same holds for FAT formatted media.
	 *
2184
	 * When this succeeds, we just return.  We have nothing
2185 2186 2187
	 * left to unlink.
	 */
	if (ret && ret != EEXIST) {
2188
	try_rename:
2189
		if (!rename(tmpfile, filename))
2190
			goto out;
2191
		ret = errno;
2192
	}
2193
	unlink_or_warn(tmpfile);
2194 2195
	if (ret) {
		if (ret != EEXIST) {
2196
			return error("unable to write sha1 filename %s: %s\n", filename, strerror(ret));
2197 2198 2199 2200
		}
		/* FIXME!!! Collision check here ? */
	}

2201
out:
2202
	if (adjust_shared_perm(filename))
2203
		return error("unable to set permission to '%s'", filename);
2204 2205 2206
	return 0;
}

L
Linus Torvalds 已提交
2207 2208
static int write_buffer(int fd, const void *buf, size_t len)
{
L
Linus Torvalds 已提交
2209
	if (write_in_full(fd, buf, len) < 0)
2210
		return error("file write error (%s)", strerror(errno));
L
Linus Torvalds 已提交
2211 2212 2213
	return 0;
}

N
Nicolas Pitre 已提交
2214
int hash_sha1_file(const void *buf, unsigned long len, const char *type,
R
Rene Scharfe 已提交
2215 2216
                   unsigned char *sha1)
{
N
Nicolas Pitre 已提交
2217
	char hdr[32];
R
Rene Scharfe 已提交
2218 2219 2220 2221 2222
	int hdrlen;
	write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
	return 0;
}

2223 2224 2225
/* Finalize a file on disk, and close it. */
static void close_sha1_file(int fd)
{
2226 2227
	if (fsync_object_files)
		fsync_or_die(fd, "sha1 file");
2228
	if (close(fd) != 0)
2229
		die_errno("error when closing sha1 file");
2230 2231
}

2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257
/* Size of directory component, including the ending '/' */
static inline int directory_size(const char *filename)
{
	const char *s = strrchr(filename, '/');
	if (!s)
		return 0;
	return s - filename + 1;
}

/*
 * This creates a temporary file in the same directory as the final
 * 'filename'
 *
 * We want to avoid cross-directory filename renames, because those
 * can have problems on various filesystems (FAT, NFS, Coda).
 */
static int create_tmpfile(char *buffer, size_t bufsiz, const char *filename)
{
	int fd, dirlen = directory_size(filename);

	if (dirlen + 20 > bufsiz) {
		errno = ENAMETOOLONG;
		return -1;
	}
	memcpy(buffer, filename, dirlen);
	strcpy(buffer + dirlen, "tmp_obj_XXXXXX");
2258
	fd = git_mkstemp_mode(buffer, 0444);
2259
	if (fd < 0 && dirlen && errno == ENOENT) {
2260
		/* Make sure the directory exists */
P
Patrick Higgins 已提交
2261
		memcpy(buffer, filename, dirlen);
2262
		buffer[dirlen-1] = 0;
2263
		if (mkdir(buffer, 0777) || adjust_shared_perm(buffer))
2264 2265 2266 2267
			return -1;

		/* Try again */
		strcpy(buffer + dirlen - 1, "/tmp_obj_XXXXXX");
2268
		fd = git_mkstemp_mode(buffer, 0444);
2269 2270 2271 2272
	}
	return fd;
}

2273 2274
static int write_loose_object(const unsigned char *sha1, char *hdr, int hdrlen,
			      void *buf, unsigned long len, time_t mtime)
2275
{
2276
	int fd, ret;
2277
	unsigned char compressed[4096];
2278
	z_stream stream;
2279 2280
	git_SHA_CTX c;
	unsigned char parano_sha1[20];
2281
	char *filename;
2282
	static char tmpfile[PATH_MAX];
2283

R
Rene Scharfe 已提交
2284
	filename = sha1_file_name(sha1);
2285
	fd = create_tmpfile(tmpfile, sizeof(tmpfile), filename);
2286 2287
	while (fd < 0 && errno == EMFILE && unuse_one_window(packed_git, -1))
		fd = create_tmpfile(tmpfile, sizeof(tmpfile), filename);
2288
	if (fd < 0) {
2289
		if (errno == EACCES)
2290 2291 2292
			return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
		else
			return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
2293 2294
	}

2295 2296
	/* Set it up */
	memset(&stream, 0, sizeof(stream));
2297
	deflateInit(&stream, zlib_compression_level);
2298
	stream.next_out = compressed;
2299
	stream.avail_out = sizeof(compressed);
2300
	git_SHA1_Init(&c);
2301 2302

	/* First header.. */
N
Nicolas Pitre 已提交
2303
	stream.next_in = (unsigned char *)hdr;
2304
	stream.avail_in = hdrlen;
2305 2306
	while (deflate(&stream, 0) == Z_OK)
		/* nothing */;
2307
	git_SHA1_Update(&c, hdr, hdrlen);
2308 2309 2310 2311

	/* Then the data itself.. */
	stream.next_in = buf;
	stream.avail_in = len;
2312
	do {
2313
		unsigned char *in0 = stream.next_in;
2314
		ret = deflate(&stream, Z_FINISH);
2315
		git_SHA1_Update(&c, in0, stream.next_in - in0);
2316 2317 2318 2319 2320 2321
		if (write_buffer(fd, compressed, stream.next_out - compressed) < 0)
			die("unable to write sha1 file");
		stream.next_out = compressed;
		stream.avail_out = sizeof(compressed);
	} while (ret == Z_OK);

2322 2323 2324 2325 2326
	if (ret != Z_STREAM_END)
		die("unable to deflate new object %s (%d)", sha1_to_hex(sha1), ret);
	ret = deflateEnd(&stream);
	if (ret != Z_OK)
		die("deflateEnd on object %s failed (%d)", sha1_to_hex(sha1), ret);
2327 2328 2329
	git_SHA1_Final(parano_sha1, &c);
	if (hashcmp(sha1, parano_sha1) != 0)
		die("confused by unstable object source data for %s", sha1_to_hex(sha1));
2330

2331
	close_sha1_file(fd);
2332

2333 2334 2335 2336 2337 2338 2339 2340 2341
	if (mtime) {
		struct utimbuf utb;
		utb.actime = mtime;
		utb.modtime = mtime;
		if (utime(tmpfile, &utb) < 0)
			warning("failed utime() on %s: %s",
				tmpfile, strerror(errno));
	}

2342
	return move_temp_to_file(tmpfile, filename);
2343
}
2344

2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368
int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
{
	unsigned char sha1[20];
	char hdr[32];
	int hdrlen;

	/* Normally if we have it in the pack then we do not bother writing
	 * it out into .git/objects/??/?{38} file.
	 */
	write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
	if (returnsha1)
		hashcpy(returnsha1, sha1);
	if (has_sha1_file(sha1))
		return 0;
	return write_loose_object(sha1, hdr, hdrlen, buf, len, 0);
}

int force_object_loose(const unsigned char *sha1, time_t mtime)
{
	void *buf;
	unsigned long len;
	enum object_type type;
	char hdr[32];
	int hdrlen;
2369
	int ret;
2370

2371
	if (has_loose_object(sha1))
2372 2373 2374 2375 2376
		return 0;
	buf = read_packed_sha1(sha1, &type, &len);
	if (!buf)
		return error("cannot read sha1_file for %s", sha1_to_hex(sha1));
	hdrlen = sprintf(hdr, "%s %lu", typename(type), len) + 1;
2377 2378 2379 2380
	ret = write_loose_object(sha1, hdr, hdrlen, buf, len, mtime);
	free(buf);

	return ret;
2381 2382
}

2383 2384 2385 2386 2387 2388 2389 2390
int has_pack_index(const unsigned char *sha1)
{
	struct stat st;
	if (stat(sha1_pack_index_name(sha1), &st))
		return 0;
	return 1;
}

2391 2392 2393 2394 2395 2396
int has_sha1_pack(const unsigned char *sha1)
{
	struct pack_entry e;
	return find_pack_entry(sha1, &e);
}

2397 2398
int has_sha1_file(const unsigned char *sha1)
{
2399 2400
	struct pack_entry e;

2401
	if (find_pack_entry(sha1, &e))
2402
		return 1;
2403
	return has_loose_object(sha1);
2404
}
J
Junio C Hamano 已提交
2405

2406 2407
static int index_mem(unsigned char *sha1, void *buf, size_t size,
		     int write_object, enum object_type type, const char *path)
2408
{
L
Linus Torvalds 已提交
2409
	int ret, re_allocated = 0;
J
Junio C Hamano 已提交
2410

2411
	if (!type)
2412
		type = OBJ_BLOB;
L
Linus Torvalds 已提交
2413 2414 2415 2416

	/*
	 * Convert blobs to git internal format
	 */
2417
	if ((type == OBJ_BLOB) && path) {
2418
		struct strbuf nbuf = STRBUF_INIT;
2419 2420
		if (convert_to_git(path, buf, size, &nbuf,
		                   write_object ? safe_crlf : 0)) {
2421
			buf = strbuf_detach(&nbuf, &size);
L
Linus Torvalds 已提交
2422 2423 2424 2425
			re_allocated = 1;
		}
	}

2426
	if (write_object)
2427
		ret = write_sha1_file(buf, size, typename(type), sha1);
R
Rene Scharfe 已提交
2428
	else
2429
		ret = hash_sha1_file(buf, size, typename(type), sha1);
2430
	if (re_allocated)
L
Linus Torvalds 已提交
2431
		free(buf);
2432 2433 2434
	return ret;
}

2435 2436
#define SMALL_FILE_SIZE (32*1024)

2437 2438 2439 2440 2441 2442 2443
int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object,
	     enum object_type type, const char *path)
{
	int ret;
	size_t size = xsize_t(st->st_size);

	if (!S_ISREG(st->st_mode)) {
2444
		struct strbuf sbuf = STRBUF_INIT;
2445 2446 2447 2448 2449 2450
		if (strbuf_read(&sbuf, fd, 4096) >= 0)
			ret = index_mem(sha1, sbuf.buf, sbuf.len, write_object,
					type, path);
		else
			ret = -1;
		strbuf_release(&sbuf);
2451 2452 2453 2454 2455 2456 2457 2458
	} else if (size <= SMALL_FILE_SIZE) {
		char *buf = xmalloc(size);
		if (size == read_in_full(fd, buf, size))
			ret = index_mem(sha1, buf, size, write_object, type,
					path);
		else
			ret = error("short read %s", strerror(errno));
		free(buf);
2459 2460 2461
	} else if (size) {
		void *buf = xmmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
		ret = index_mem(sha1, buf, size, write_object, type, path);
2462
		munmap(buf, size);
2463 2464 2465
	} else
		ret = index_mem(sha1, NULL, size, write_object, type, path);
	close(fd);
2466
	return ret;
J
Junio C Hamano 已提交
2467
}
2468 2469 2470 2471

int index_path(unsigned char *sha1, const char *path, struct stat *st, int write_object)
{
	int fd;
2472
	struct strbuf sb = STRBUF_INIT;
2473 2474 2475 2476 2477 2478 2479

	switch (st->st_mode & S_IFMT) {
	case S_IFREG:
		fd = open(path, O_RDONLY);
		if (fd < 0)
			return error("open(\"%s\"): %s", path,
				     strerror(errno));
2480
		if (index_fd(sha1, fd, st, write_object, OBJ_BLOB, path) < 0)
2481 2482 2483 2484
			return error("%s: failed to insert into database",
				     path);
		break;
	case S_IFLNK:
2485
		if (strbuf_readlink(&sb, path, st->st_size)) {
2486 2487 2488 2489
			char *errstr = strerror(errno);
			return error("readlink(\"%s\"): %s", path,
			             errstr);
		}
R
Rene Scharfe 已提交
2490
		if (!write_object)
2491 2492
			hash_sha1_file(sb.buf, sb.len, blob_type, sha1);
		else if (write_sha1_file(sb.buf, sb.len, blob_type, sha1))
2493 2494
			return error("%s: failed to insert into database",
				     path);
2495
		strbuf_release(&sb);
2496
		break;
2497 2498
	case S_IFDIR:
		return resolve_gitlink_ref(path, "HEAD", sha1);
2499 2500 2501 2502 2503
	default:
		return error("%s: unsupported file type", path);
	}
	return 0;
}
2504 2505 2506

int read_pack_header(int fd, struct pack_header *header)
{
2507 2508 2509 2510
	if (read_in_full(fd, header, sizeof(*header)) < sizeof(*header))
		/* "eof before pack header was fully read" */
		return PH_ERROR_EOF;

2511 2512 2513 2514 2515 2516 2517 2518
	if (header->hdr_signature != htonl(PACK_SIGNATURE))
		/* "protocol error (pack signature mismatch detected)" */
		return PH_ERROR_PACK_SIGNATURE;
	if (!pack_version_ok(header->hdr_version))
		/* "protocol error (pack version unsupported)" */
		return PH_ERROR_PROTOCOL;
	return 0;
}