op public repos

Blob

Date:: Mon Feb 11 13:20:09 2019 UTC
Message:: don't close input files in dump_delta_chain_to_file(); caller does it
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/uio.h>
21 #include <sys/mman.h>
22 
23 #include <fcntl.h>
24 #include <errno.h>
25 #include <stdio.h>
26 #include <stdint.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <limits.h>
30 #include <sha1.h>
31 #include <endian.h>
32 #include <zlib.h>
33 #include <imsg.h>
34 
35 #include "got_error.h"
36 #include "got_object.h"
37 #include "got_opentemp.h"
38 
39 #include "got_lib_sha1.h"
40 #include "got_lib_path.h"
41 #include "got_lib_delta.h"
42 #include "got_lib_inflate.h"
43 #include "got_lib_object.h"
44 #include "got_lib_privsep.h"
45 #include "got_lib_pack.h"
46 
47 #ifndef nitems
48 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
49 #endif
50 
51 #ifndef MIN
52 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
53 #endif
54 
55 static const struct got_error *
56 verify_fanout_table(uint32_t *fanout_table)
57 {
58 	int i;
59 
60 	for (i = 0; i < 0xff - 1; i++) {
61 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
62 			return got_error(GOT_ERR_BAD_PACKIDX);
63 	}
64 
65 	return NULL;
66 }
67 
68 const struct got_error *
69 got_pack_get_packfile_size(size_t *size, const char *path)
70 {
71 	struct stat sb;
72 	char *dot;
73 
74 	*size = 0;
75 
76 	dot = strrchr(path, '.');
77 	if (dot == NULL)
78 		return got_error(GOT_ERR_BAD_PATH);
79 
80 	/* Path must point to a pack index or to a pack file. */
81 	if (strcmp(dot, GOT_PACKIDX_SUFFIX) == 0) {
82 		const struct got_error *err = NULL;
83 		char *path_pack;
84 		char base_path[PATH_MAX];
85 
86 		/* Convert pack index path to pack file path. */
87 		if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
88 			return got_error(GOT_ERR_NO_SPACE);
89 		dot = strrchr(base_path, '.');
90 		if (dot == NULL)
91 			return got_error(GOT_ERR_BAD_PATH);
92 		*dot = '\0';
93 		if (asprintf(&path_pack, "%s.pack", base_path) == -1)
94 			return got_error_from_errno();
95 
96 		if (stat(path_pack, &sb) != 0)
97 			err = got_error_from_errno();
98 		free(path_pack);
99 		if (err)
100 			return err;
101 	} else if (strcmp(dot, GOT_PACKFILE_SUFFIX) == 0) {
102 		if (stat(path, &sb) != 0)
103 			return got_error_from_errno();
104 	} else
105 		return got_error(GOT_ERR_BAD_PATH);
106 
107 	*size = sb.st_size;
108 	return 0;
109 }
110 
111 const struct got_error *
112 got_packidx_init_hdr(struct got_packidx *p, int verify)
113 {
114 	const struct got_error *err = NULL;
115 	struct got_packidx_v2_hdr *h;
116 	SHA1_CTX ctx;
117 	uint8_t sha1[SHA1_DIGEST_LENGTH];
118 	size_t nobj, len_fanout, len_ids, offset, remain;
119 	ssize_t n;
120 
121 	SHA1Init(&ctx);
122 
123 	h = &p->hdr;
124 	offset = 0;
125 	remain = p->len;
126 
127 	if (remain < sizeof(*h->magic)) {
128 		err = got_error(GOT_ERR_BAD_PACKIDX);
129 		goto done;
130 	}
131 	if (p->map)
132 		h->magic = (uint32_t *)(p->map + offset);
133 	else {
134 		h->magic = malloc(sizeof(*h->magic));
135 		if (h->magic == NULL) {
136 			err = got_error_from_errno();
137 			goto done;
138 		}
139 		n = read(p->fd, h->magic, sizeof(*h->magic));
140 		if (n < 0)
141 			err = got_error_from_errno();
142 		else if (n != sizeof(*h->magic)) {
143 			err = got_error(GOT_ERR_BAD_PACKIDX);
144 			goto done;
145 		}
146 	}
147 	if (betoh32(*h->magic) != GOT_PACKIDX_V2_MAGIC) {
148 		err = got_error(GOT_ERR_BAD_PACKIDX);
149 		goto done;
150 	}
151 	offset += sizeof(*h->magic);
152 	remain -= sizeof(*h->magic);
153 
154 	if (verify)
155 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
156 
157 	if (remain < sizeof(*h->version)) {
158 		err = got_error(GOT_ERR_BAD_PACKIDX);
159 		goto done;
160 	}
161 	if (p->map)
162 		h->version = (uint32_t *)(p->map + offset);
163 	else {
164 		h->version = malloc(sizeof(*h->version));
165 		if (h->version == NULL) {
166 			err = got_error_from_errno();
167 			goto done;
168 		}
169 		n = read(p->fd, h->version, sizeof(*h->version));
170 		if (n < 0)
171 			err = got_error_from_errno();
172 		else if (n != sizeof(*h->version)) {
173 			err = got_error(GOT_ERR_BAD_PACKIDX);
174 			goto done;
175 		}
176 	}
177 	if (betoh32(*h->version) != GOT_PACKIDX_VERSION) {
178 		err = got_error(GOT_ERR_BAD_PACKIDX);
179 		goto done;
180 	}
181 	offset += sizeof(*h->version);
182 	remain -= sizeof(*h->version);
183 
184 	if (verify)
185 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
186 
187 	len_fanout =
188 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
189 	if (remain < len_fanout) {
190 		err = got_error(GOT_ERR_BAD_PACKIDX);
191 		goto done;
192 	}
193 	if (p->map)
194 		h->fanout_table = (uint32_t *)(p->map + offset);
195 	else {
196 		h->fanout_table = malloc(len_fanout);
197 		if (h->fanout_table == NULL) {
198 			err = got_error_from_errno();
199 			goto done;
200 		}
201 		n = read(p->fd, h->fanout_table, len_fanout);
202 		if (n < 0)
203 			err = got_error_from_errno();
204 		else if (n != len_fanout) {
205 			err = got_error(GOT_ERR_BAD_PACKIDX);
206 			goto done;
207 		}
208 	}
209 	err = verify_fanout_table(h->fanout_table);
210 	if (err)
211 		goto done;
212 	if (verify)
213 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
214 	offset += len_fanout;
215 	remain -= len_fanout;
216 
217 	nobj = betoh32(h->fanout_table[0xff]);
218 	len_ids = nobj * sizeof(*h->sorted_ids);
219 	if (len_ids <= nobj || len_ids > remain) {
220 		err = got_error(GOT_ERR_BAD_PACKIDX);
221 		goto done;
222 	}
223 	if (p->map)
224 		h->sorted_ids =
225 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
226 	else {
227 		h->sorted_ids = malloc(len_ids);
228 		if (h->sorted_ids == NULL) {
229 			err = got_error(GOT_ERR_BAD_PACKIDX);
230 			goto done;
231 		}
232 		n = read(p->fd, h->sorted_ids, len_ids);
233 		if (n < 0)
234 			err = got_error_from_errno();
235 		else if (n != len_ids) {
236 			err = got_error(GOT_ERR_BAD_PACKIDX);
237 			goto done;
238 		}
239 	}
240 	if (verify)
241 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
242 	offset += len_ids;
243 	remain -= len_ids;
244 
245 	if (remain < nobj * sizeof(*h->crc32)) {
246 		err = got_error(GOT_ERR_BAD_PACKIDX);
247 		goto done;
248 	}
249 	if (p->map)
250 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
251 	else {
252 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
253 		if (h->crc32 == NULL) {
254 			err = got_error_from_errno();
255 			goto done;
256 		}
257 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
258 		if (n < 0)
259 			err = got_error_from_errno();
260 		else if (n != nobj * sizeof(*h->crc32)) {
261 			err = got_error(GOT_ERR_BAD_PACKIDX);
262 			goto done;
263 		}
264 	}
265 	if (verify)
266 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
267 	remain -= nobj * sizeof(*h->crc32);
268 	offset += nobj * sizeof(*h->crc32);
269 
270 	if (remain < nobj * sizeof(*h->offsets)) {
271 		err = got_error(GOT_ERR_BAD_PACKIDX);
272 		goto done;
273 	}
274 	if (p->map)
275 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
276 	else {
277 		h->offsets = malloc(nobj * sizeof(*h->offsets));
278 		if (h->offsets == NULL) {
279 			err = got_error_from_errno();
280 			goto done;
281 		}
282 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
283 		if (n < 0)
284 			err = got_error_from_errno();
285 		else if (n != nobj * sizeof(*h->offsets)) {
286 			err = got_error(GOT_ERR_BAD_PACKIDX);
287 			goto done;
288 		}
289 	}
290 	if (verify)
291 		SHA1Update(&ctx, (uint8_t *)h->offsets,
292 		    nobj * sizeof(*h->offsets));
293 	remain -= nobj * sizeof(*h->offsets);
294 	offset += nobj * sizeof(*h->offsets);
295 
296 	/* Large file offsets are contained only in files > 2GB. */
297 	if (p->len <= 0x80000000)
298 		goto checksum;
299 
300 	if (remain < nobj * sizeof(*h->large_offsets)) {
301 		err = got_error(GOT_ERR_BAD_PACKIDX);
302 		goto done;
303 	}
304 	if (p->map)
305 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
306 	else {
307 		h->offsets = malloc(nobj * sizeof(*h->large_offsets));
308 		if (h->offsets == NULL) {
309 			err = got_error_from_errno();
310 			goto done;
311 		}
312 		n = read(p->fd, h->large_offsets,
313 		    nobj * sizeof(*h->large_offsets));
314 		if (n < 0)
315 			err = got_error_from_errno();
316 		else if (n != nobj * sizeof(*h->large_offsets)) {
317 			err = got_error(GOT_ERR_BAD_PACKIDX);
318 			goto done;
319 		}
320 	}
321 	if (verify)
322 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
323 		    nobj * sizeof(*h->large_offsets));
324 	remain -= nobj * sizeof(*h->large_offsets);
325 	offset += nobj * sizeof(*h->large_offsets);
326 
327 checksum:
328 	if (remain < sizeof(*h->trailer)) {
329 		err = got_error(GOT_ERR_BAD_PACKIDX);
330 		goto done;
331 	}
332 	if (p->map)
333 		h->trailer =
334 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
335 	else {
336 		h->trailer = malloc(sizeof(*h->trailer));
337 		if (h->trailer == NULL) {
338 			err = got_error_from_errno();
339 			goto done;
340 		}
341 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
342 		if (n < 0)
343 			err = got_error_from_errno();
344 		else if (n != sizeof(*h->trailer)) {
345 			err = got_error(GOT_ERR_BAD_PACKIDX);
346 			goto done;
347 		}
348 	}
349 	if (verify) {
350 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
351 		SHA1Final(sha1, &ctx);
352 		if (memcmp(h->trailer->packidx_sha1, sha1,
353 		    SHA1_DIGEST_LENGTH) != 0)
354 			err = got_error(GOT_ERR_PACKIDX_CSUM);
355 	}
356 done:
357 	return err;
358 }
359 
360 const struct got_error *
361 got_packidx_open(struct got_packidx **packidx, const char *path, int verify)
362 {
363 	const struct got_error *err = NULL;
364 	struct got_packidx *p;
365 
366 	*packidx = NULL;
367 
368 	p = calloc(1, sizeof(*p));
369 	if (p == NULL)
370 		return got_error_from_errno();
371 
372 	p->fd = open(path, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
373 	if (p->fd == -1)
374 		return got_error_from_errno();
375 
376 	err = got_pack_get_packfile_size(&p->len, path);
377 	if (err) {
378 		close(p->fd);
379 		free(p);
380 		return err;
381 	}
382 	if (p->len < sizeof(p->hdr)) {
383 		err = got_error(GOT_ERR_BAD_PACKIDX);
384 		close(p->fd);
385 		free(p);
386 		return err;
387 	}
388 
389 	p->path_packidx = strdup(path);
390 	if (p->path_packidx == NULL) {
391 		err = got_error_from_errno();
392 		goto done;
393 	}
394 
395 #ifndef GOT_PACK_NO_MMAP
396 	p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
397 	if (p->map == MAP_FAILED)
398 		p->map = NULL; /* fall back to read(2) */
399 #endif
400 
401 	err = got_packidx_init_hdr(p, verify);
402 done:
403 	if (err)
404 		got_packidx_close(p);
405 	else
406 		*packidx = p;
407 
408 	return err;
409 }
410 
411 const struct got_error *
412 got_packidx_close(struct got_packidx *packidx)
413 {
414 	const struct got_error *err = NULL;
415 
416 	free(packidx->path_packidx);
417 	if (packidx->map) {
418 		if (munmap(packidx->map, packidx->len) == -1)
419 			err = got_error_from_errno();
420 	} else {
421 		free(packidx->hdr.magic);
422 		free(packidx->hdr.version);
423 		free(packidx->hdr.fanout_table);
424 		free(packidx->hdr.sorted_ids);
425 		free(packidx->hdr.crc32);
426 		free(packidx->hdr.offsets);
427 		free(packidx->hdr.large_offsets);
428 		free(packidx->hdr.trailer);
429 	}
430 	if (close(packidx->fd) != 0 && err == NULL)
431 		err = got_error_from_errno();
432 	free(packidx);
433 
434 	return err;
435 }
436 
437 static off_t
438 get_object_offset(struct got_packidx *packidx, int idx)
439 {
440 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
441 	uint32_t offset = betoh32(packidx->hdr.offsets[idx]);
442 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
443 		uint64_t loffset;
444 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
445 		if (idx < 0 || idx > totobj ||
446 		    packidx->hdr.large_offsets == NULL)
447 			return -1;
448 		loffset = betoh64(packidx->hdr.large_offsets[idx]);
449 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
450 	}
451 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
452 }
453 
454 int
455 got_packidx_get_object_idx(struct got_packidx *packidx, struct got_object_id *id)
456 {
457 	u_int8_t id0 = id->sha1[0];
458 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
459 	int left = 0, right = totobj - 1;
460 
461 	if (id0 > 0)
462 		left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
463 
464 	while (left <= right) {
465 		struct got_packidx_object_id *oid;
466 		int i, cmp;
467 
468 		i = ((left + right) / 2);
469 		oid = &packidx->hdr.sorted_ids[i];
470 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
471 		if (cmp == 0)
472 			return i;
473 		else if (cmp > 0)
474 			left = i + 1;
475 		else if (cmp < 0)
476 			right = i - 1;
477 	}
478 
479 	return -1;
480 }
481 
482 const struct got_error *
483 got_pack_stop_privsep_child(struct got_pack *pack)
484 {
485 	const struct got_error *err = NULL;
486 
487 	if (pack->privsep_child == NULL)
488 		return NULL;
489 
490 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
491 	if (err)
492 		return err;
493 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
494 	free(pack->privsep_child);
495 	pack->privsep_child = NULL;
496 	return err;
497 }
498 
499 const struct got_error *
500 got_pack_close(struct got_pack *pack)
501 {
502 	const struct got_error *err = NULL;
503 
504 	err = got_pack_stop_privsep_child(pack);
505 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
506 		err = got_error_from_errno();
507 	if (pack->fd != -1 && close(pack->fd) != 0 && err == NULL)
508 		err = got_error_from_errno();
509 	pack->fd = -1;
510 	free(pack->path_packfile);
511 	pack->path_packfile = NULL;
512 	pack->filesize = 0;
513 
514 	return err;
515 }
516 
517 static const struct got_error *
518 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
519     struct got_pack *pack, off_t offset)
520 {
521 	uint8_t t = 0;
522 	uint64_t s = 0;
523 	uint8_t sizeN;
524 	size_t mapoff = 0;
525 	int i = 0;
526 
527 	*len = 0;
528 
529 	if (offset >= pack->filesize)
530 		return got_error(GOT_ERR_PACK_OFFSET);
531 
532 	if (pack->map) {
533 		mapoff = (size_t)offset;
534 	} else {
535 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
536 			return got_error_from_errno();
537 	}
538 
539 	do {
540 		/* We do not support size values which don't fit in 64 bit. */
541 		if (i > 9)
542 			return got_error(GOT_ERR_NO_SPACE);
543 
544 		if (pack->map) {
545 			sizeN = *(pack->map + mapoff);
546 			mapoff += sizeof(sizeN);
547 		} else {
548 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
549 			if (n < 0)
550 				return got_error_from_errno();
551 			if (n != sizeof(sizeN))
552 				return got_error(GOT_ERR_BAD_PACKFILE);
553 		}
554 		*len += sizeof(sizeN);
555 
556 		if (i == 0) {
557 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
558 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
559 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
560 		} else {
561 			size_t shift = 4 + 7 * (i - 1);
562 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
563 		}
564 		i++;
565 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
566 
567 	*type = t;
568 	*size = s;
569 	return NULL;
570 }
571 
572 static const struct got_error *
573 open_plain_object(struct got_object **obj, const char *path_packfile,
574     struct got_object_id *id, uint8_t type, off_t offset, size_t size, int idx)
575 {
576 	*obj = calloc(1, sizeof(**obj));
577 	if (*obj == NULL)
578 		return got_error_from_errno();
579 
580 	(*obj)->path_packfile = strdup(path_packfile);
581 	if ((*obj)->path_packfile == NULL) {
582 		const struct got_error *err = got_error_from_errno();
583 		free(*obj);
584 		*obj = NULL;
585 		return err;
586 	}
587 
588 	(*obj)->type = type;
589 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
590 	(*obj)->pack_idx = idx;
591 	(*obj)->hdrlen = 0;
592 	(*obj)->size = size;
593 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
594 	(*obj)->pack_offset = offset;
595 
596 	return NULL;
597 }
598 
599 static const struct got_error *
600 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
601     off_t delta_offset)
602 {
603 	int64_t o = 0;
604 	uint8_t offN;
605 	int i = 0;
606 
607 	*len = 0;
608 
609 	do {
610 		/* We do not support offset values which don't fit in 64 bit. */
611 		if (i > 8)
612 			return got_error(GOT_ERR_NO_SPACE);
613 
614 		if (pack->map) {
615 			size_t mapoff;
616 			if (delta_offset >= pack->filesize)
617 				return got_error(GOT_ERR_PACK_OFFSET);
618 			mapoff = (size_t)delta_offset + *len;
619 			offN = *(pack->map + mapoff);
620 		} else {
621 			ssize_t n;
622 			n = read(pack->fd, &offN, sizeof(offN));
623 			if (n < 0)
624 				return got_error_from_errno();
625 			if (n != sizeof(offN))
626 				return got_error(GOT_ERR_BAD_PACKFILE);
627 		}
628 		*len += sizeof(offN);
629 
630 		if (i == 0)
631 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
632 		else {
633 			o++;
634 			o <<= 7;
635 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
636 		}
637 		i++;
638 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
639 
640 	*offset = o;
641 	return NULL;
642 }
643 
644 static const struct got_error *
645 parse_offset_delta(off_t *base_offset, size_t *len, struct got_pack *pack,
646     off_t offset, int tslen)
647 {
648 	const struct got_error *err;
649 	int64_t negoffset;
650 	size_t negofflen;
651 
652 	*len = 0;
653 
654 	err = parse_negative_offset(&negoffset, &negofflen, pack,
655 	    offset + tslen);
656 	if (err)
657 		return err;
658 
659 	/* Compute the base object's offset (must be in the same pack file). */
660 	*base_offset = (offset - negoffset);
661 	if (*base_offset <= 0)
662 		return got_error(GOT_ERR_BAD_PACKFILE);
663 
664 	*len = negofflen;
665 	return NULL;
666 }
667 
668 static const struct got_error *
669 resolve_delta_chain(struct got_delta_chain *, struct got_packidx *,
670     struct got_pack *, off_t, size_t, int, size_t, unsigned int);
671 
672 static const struct got_error *
673 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
674     int delta_type, size_t delta_size, size_t delta_data_offset,
675     uint8_t *delta_buf, size_t delta_len)
676 {
677 	struct got_delta *delta;
678 
679 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
680 	    delta_data_offset, delta_buf,
681 	    delta_len);
682 	if (delta == NULL)
683 		return got_error_from_errno();
684 	/* delta is freed in got_object_close() */
685 	deltas->nentries++;
686 	SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
687 	return NULL;
688 }
689 
690 static const struct got_error *
691 resolve_offset_delta(struct got_delta_chain *deltas,
692     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
693     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
694 
695 {
696 	const struct got_error *err;
697 	off_t base_offset;
698 	uint8_t base_type;
699 	uint64_t base_size;
700 	size_t base_tslen;
701 	off_t delta_data_offset;
702 	uint8_t *delta_buf;
703 	size_t delta_len, consumed;
704 
705 	err = parse_offset_delta(&base_offset, &consumed, pack,
706 	    delta_offset, tslen);
707 	if (err)
708 		return err;
709 
710 	delta_data_offset = delta_offset + tslen + consumed;
711 	if (delta_data_offset >= pack->filesize)
712 		return got_error(GOT_ERR_PACK_OFFSET);
713 
714 	if (pack->map == NULL) {
715 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
716 		if (delta_data_offset == -1)
717 			return got_error_from_errno();
718 	}
719 
720 	if (pack->map) {
721 		size_t mapoff = (size_t)delta_data_offset;
722 		err = got_inflate_to_mem_mmap(&delta_buf, &delta_len, pack->map,
723 		    mapoff, pack->filesize - mapoff);
724 		if (err)
725 			return err;
726 	} else {
727 
728 		err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
729 		if (err)
730 			return err;
731 	}
732 
733 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
734 	    delta_data_offset, delta_buf, delta_len);
735 	if (err)
736 		goto done;
737 
738 	/* An offset delta must be in the same packfile. */
739 	if (base_offset >= pack->filesize) {
740 		err = got_error(GOT_ERR_PACK_OFFSET);
741 		goto done;
742 	}
743 
744 	err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
745 	    pack, base_offset);
746 	if (err)
747 		goto done;
748 
749 	err = resolve_delta_chain(deltas, packidx, pack, base_offset,
750 	    base_tslen, base_type, base_size, recursion - 1);
751 done:
752 	if (err)
753 		free(delta_buf);
754 	return err;
755 }
756 
757 static const struct got_error *
758 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
759     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
760     size_t delta_size, unsigned int recursion)
761 {
762 	const struct got_error *err;
763 	struct got_object_id id;
764 	int idx;
765 	off_t base_offset;
766 	uint8_t base_type;
767 	uint64_t base_size;
768 	size_t base_tslen;
769 	off_t delta_data_offset;
770 	uint8_t *delta_buf = NULL;
771 	size_t delta_len;
772 
773 	if (delta_offset >= pack->filesize)
774 		return got_error(GOT_ERR_PACK_OFFSET);
775 	delta_data_offset = delta_offset + tslen;
776 	if (delta_data_offset >= pack->filesize)
777 		return got_error(GOT_ERR_PACK_OFFSET);
778 
779 	if (pack->map == NULL) {
780 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
781 		if (delta_data_offset == -1)
782 			return got_error_from_errno();
783 	}
784 
785 
786 	if (pack->map) {
787 		size_t mapoff = (size_t)delta_data_offset;
788 		memcpy(&id, pack->map + mapoff, sizeof(id));
789 		mapoff += sizeof(id);
790 		err = got_inflate_to_mem_mmap(&delta_buf, &delta_len, pack->map,
791 		    mapoff, pack->filesize - mapoff);
792 		if (err)
793 			goto done;
794 	} else {
795 		ssize_t n = read(pack->fd, &id, sizeof(id));
796 		if (n < 0)
797 			return got_error_from_errno();
798 		if (n != sizeof(id))
799 			return got_error(GOT_ERR_BAD_PACKFILE);
800 		err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
801 		if (err)
802 			goto done;
803 	}
804 
805 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
806 	    delta_data_offset, delta_buf, delta_len);
807 	if (err)
808 		goto done;
809 
810 	/* Delta base must be in the same pack file. */
811 	idx = got_packidx_get_object_idx(packidx, &id);
812 	if (idx == -1) {
813 		err = got_error(GOT_ERR_BAD_PACKFILE);
814 		goto done;
815 	}
816 
817 	base_offset = get_object_offset(packidx, idx);
818 	if (base_offset == (uint64_t)-1) {
819 		err = got_error(GOT_ERR_BAD_PACKIDX);
820 		goto done;
821 	}
822 
823 	if (base_offset >= pack->filesize) {
824 		err = got_error(GOT_ERR_PACK_OFFSET);
825 		goto done;
826 	}
827 
828 	err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
829 	    pack, base_offset);
830 	if (err)
831 		goto done;
832 
833 	err = resolve_delta_chain(deltas, packidx, pack, base_offset,
834 	    base_tslen, base_type, base_size, recursion - 1);
835 done:
836 	if (err)
837 		free(delta_buf);
838 	return err;
839 }
840 
841 static const struct got_error *
842 resolve_delta_chain(struct got_delta_chain *deltas, struct got_packidx *packidx,
843     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
844     size_t delta_size, unsigned int recursion)
845 {
846 	const struct got_error *err = NULL;
847 
848 	if (--recursion == 0)
849 		return got_error(GOT_ERR_RECURSION);
850 
851 	switch (delta_type) {
852 	case GOT_OBJ_TYPE_COMMIT:
853 	case GOT_OBJ_TYPE_TREE:
854 	case GOT_OBJ_TYPE_BLOB:
855 	case GOT_OBJ_TYPE_TAG:
856 		/* Plain types are the final delta base. Recursion ends. */
857 		err = add_delta(deltas, delta_offset, tslen, delta_type,
858 		    delta_size, 0, NULL, 0);
859 		break;
860 	case GOT_OBJ_TYPE_OFFSET_DELTA:
861 		err = resolve_offset_delta(deltas, packidx, pack,
862 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
863 		break;
864 	case GOT_OBJ_TYPE_REF_DELTA:
865 		err = resolve_ref_delta(deltas, packidx, pack,
866 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
867 		break;
868 	default:
869 		return got_error(GOT_ERR_OBJ_TYPE);
870 	}
871 
872 	return err;
873 }
874 
875 static const struct got_error *
876 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
877     struct got_pack *pack, struct got_object_id *id, off_t offset,
878     size_t tslen, int delta_type, size_t delta_size, int idx)
879 {
880 	const struct got_error *err = NULL;
881 	int resolved_type;
882 
883 	*obj = calloc(1, sizeof(**obj));
884 	if (*obj == NULL)
885 		return got_error_from_errno();
886 
887 	(*obj)->flags = 0;
888 	(*obj)->hdrlen = 0;
889 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
890 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
891 	(*obj)->pack_offset = offset + tslen;
892 
893 	(*obj)->path_packfile = strdup(pack->path_packfile);
894 	if ((*obj)->path_packfile == NULL) {
895 		err = got_error_from_errno();
896 		goto done;
897 	}
898 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
899 	(*obj)->pack_idx = idx;
900 
901 	SIMPLEQ_INIT(&(*obj)->deltas.entries);
902 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
903 
904 	err = resolve_delta_chain(&(*obj)->deltas, packidx, pack, offset,
905 	    tslen, delta_type, delta_size, GOT_DELTA_CHAIN_RECURSION_MAX);
906 	if (err)
907 		goto done;
908 
909 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
910 	if (err)
911 		goto done;
912 	(*obj)->type = resolved_type;
913 done:
914 	if (err) {
915 		got_object_close(*obj);
916 		*obj = NULL;
917 	}
918 	return err;
919 }
920 
921 const struct got_error *
922 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
923     struct got_packidx *packidx, int idx, struct got_object_id *id)
924 {
925 	const struct got_error *err = NULL;
926 	off_t offset;
927 	uint8_t type;
928 	uint64_t size;
929 	size_t tslen;
930 
931 	*obj = NULL;
932 
933 	offset = get_object_offset(packidx, idx);
934 	if (offset == (uint64_t)-1)
935 		return got_error(GOT_ERR_BAD_PACKIDX);
936 
937 	err = parse_object_type_and_size(&type, &size, &tslen, pack, offset);
938 	if (err)
939 		return err;
940 
941 	switch (type) {
942 	case GOT_OBJ_TYPE_COMMIT:
943 	case GOT_OBJ_TYPE_TREE:
944 	case GOT_OBJ_TYPE_BLOB:
945 	case GOT_OBJ_TYPE_TAG:
946 		err = open_plain_object(obj, pack->path_packfile, id, type,
947 		    offset + tslen, size, idx);
948 		break;
949 	case GOT_OBJ_TYPE_OFFSET_DELTA:
950 	case GOT_OBJ_TYPE_REF_DELTA:
951 		err = open_delta_object(obj, packidx, pack, id, offset,
952 		    tslen, type, size, idx);
953 		break;
954 	default:
955 		err = got_error(GOT_ERR_OBJ_TYPE);
956 		break;
957 	}
958 
959 	return err;
960 }
961 
962 static const struct got_error *
963 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
964 {
965 	struct got_delta *delta;
966 	uint64_t base_size = 0, result_size = 0;
967 
968 	*max_size = 0;
969 	SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
970 		/* Plain object types are the delta base. */
971 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
972 		    delta->type != GOT_OBJ_TYPE_TREE &&
973 		    delta->type != GOT_OBJ_TYPE_BLOB &&
974 		    delta->type != GOT_OBJ_TYPE_TAG) {
975 			const struct got_error *err;
976 			err = got_delta_get_sizes(&base_size, &result_size,
977 			    delta->delta_buf, delta->delta_len);
978 			if (err)
979 				return err;
980 		} else
981 			base_size = delta->size;
982 		if (base_size > *max_size)
983 			*max_size = base_size;
984 		if (result_size > *max_size)
985 			*max_size = result_size;
986 	}
987 
988 	return NULL;
989 }
990 
991 const struct got_error *
992 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj)
993 {
994 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
995 		return got_error(GOT_ERR_OBJ_TYPE);
996 
997 	return get_delta_chain_max_size(size, &obj->deltas);
998 }
999 
1000 static const struct got_error *
1001 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1002     struct got_pack *pack, FILE *outfile, FILE *base_file, FILE *accum_file)
1003 {
1004 	const struct got_error *err = NULL;
1005 	struct got_delta *delta;
1006 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1007 	size_t base_bufsz = 0, accum_size = 0;
1008 	uint64_t max_size;
1009 	int n = 0;
1010 
1011 	*result_size = 0;
1012 
1013 	if (SIMPLEQ_EMPTY(&deltas->entries))
1014 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1015 
1016 	/* We process small enough files entirely in memory for speed. */
1017 	err = get_delta_chain_max_size(&max_size, deltas);
1018 	if (err)
1019 		return err;
1020 	if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1021 		accum_buf = malloc(max_size);
1022 		if (accum_buf == NULL)
1023 			return got_error_from_errno();
1024 		base_file = NULL;
1025 		accum_file = NULL;
1026 	}
1027 
1028 	/* Deltas are ordered in ascending order. */
1029 	SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1030 		if (n == 0) {
1031 			size_t mapoff;
1032 			off_t delta_data_offset;
1033 
1034 			/* Plain object types are the delta base. */
1035 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1036 			    delta->type != GOT_OBJ_TYPE_TREE &&
1037 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1038 			    delta->type != GOT_OBJ_TYPE_TAG) {
1039 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1040 				goto done;
1041 			}
1042 
1043 			delta_data_offset = delta->offset + delta->tslen;
1044 			if (delta_data_offset >= pack->filesize) {
1045 				err = got_error(GOT_ERR_PACK_OFFSET);
1046 				goto done;
1047 			}
1048 			if (pack->map == NULL) {
1049 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1050 				    == -1) {
1051 					err = got_error_from_errno();
1052 					goto done;
1053 				}
1054 			}
1055 			if (base_file) {
1056 				if (pack->map) {
1057 					mapoff = (size_t)delta_data_offset;
1058 					err = got_inflate_to_file_mmap(
1059 					    &base_bufsz, pack->map, mapoff,
1060 					    pack->filesize - mapoff, base_file);
1061 				} else
1062 					err = got_inflate_to_file_fd(
1063 					    &base_bufsz, pack->fd, base_file);
1064 			} else {
1065 				if (pack->map) {
1066 					mapoff = (size_t)delta_data_offset;
1067 					err = got_inflate_to_mem_mmap(&base_buf,
1068 					    &base_bufsz, pack->map, mapoff,
1069 					    pack->filesize - mapoff);
1070 				} else
1071 					err = got_inflate_to_mem_fd(&base_buf,
1072 					    &base_bufsz, pack->fd);
1073 			}
1074 			if (err)
1075 				goto done;
1076 			n++;
1077 			if (base_file)
1078 				rewind(base_file);
1079 			continue;
1080 		}
1081 
1082 		if (base_buf) {
1083 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1084 			    delta->delta_buf, delta->delta_len, accum_buf,
1085 			    &accum_size, max_size);
1086 			n++;
1087 		} else {
1088 			err = got_delta_apply(base_file, delta->delta_buf,
1089 			    delta->delta_len,
1090 			    /* Final delta application writes to output file. */
1091 			    ++n < deltas->nentries ? accum_file : outfile,
1092 			    &accum_size);
1093 		}
1094 		if (err)
1095 			goto done;
1096 
1097 		if (n < deltas->nentries) {
1098 			/* Accumulated delta becomes the new base. */
1099 			if (base_buf) {
1100 				uint8_t *tmp = accum_buf;
1101 				/*
1102 				 * Base buffer switches roles with accumulation
1103 				 * buffer. Ensure it can hold the largest
1104 				 * result in the delta chain. The initial
1105 				 * allocation might have been smaller.
1106 				 */
1107 				if (base_bufsz < max_size) {
1108 					uint8_t *p;
1109 					p = reallocarray(base_buf, 1, max_size);
1110 					if (p == NULL) {
1111 						err = got_error_from_errno();
1112 						goto done;
1113 					}
1114 					base_buf = p;
1115 					base_bufsz = max_size;
1116 				}
1117 				accum_buf = base_buf;
1118 				base_buf = tmp;
1119 			} else {
1120 				FILE *tmp = accum_file;
1121 				accum_file = base_file;
1122 				base_file = tmp;
1123 				rewind(base_file);
1124 				rewind(accum_file);
1125 			}
1126 		}
1127 	}
1128 
1129 done:
1130 	free(base_buf);
1131 	if (accum_buf) {
1132 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1133 		free(accum_buf);
1134 		if (len != accum_size)
1135 			err = got_ferror(outfile, GOT_ERR_IO);
1136 	}
1137 	rewind(outfile);
1138 	if (err == NULL)
1139 		*result_size = accum_size;
1140 	return err;
1141 }
1142 
1143 static const struct got_error *
1144 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1145     struct got_delta_chain *deltas, struct got_pack *pack)
1146 {
1147 	const struct got_error *err = NULL;
1148 	struct got_delta *delta;
1149 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1150 	size_t base_bufsz = 0, accum_size = 0;
1151 	uint64_t max_size;
1152 	int n = 0;
1153 
1154 	*outbuf = NULL;
1155 	*outlen = 0;
1156 
1157 	if (SIMPLEQ_EMPTY(&deltas->entries))
1158 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1159 
1160 	err = get_delta_chain_max_size(&max_size, deltas);
1161 	if (err)
1162 		return err;
1163 	accum_buf = malloc(max_size);
1164 	if (accum_buf == NULL)
1165 		return got_error_from_errno();
1166 
1167 	/* Deltas are ordered in ascending order. */
1168 	SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1169 		if (n == 0) {
1170 			size_t delta_data_offset;
1171 
1172 			/* Plain object types are the delta base. */
1173 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1174 			    delta->type != GOT_OBJ_TYPE_TREE &&
1175 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1176 			    delta->type != GOT_OBJ_TYPE_TAG) {
1177 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1178 				goto done;
1179 			}
1180 
1181 			delta_data_offset = delta->offset + delta->tslen;
1182 			if (delta_data_offset >= pack->filesize) {
1183 				err = got_error(GOT_ERR_PACK_OFFSET);
1184 				goto done;
1185 			}
1186 			if (pack->map) {
1187 				size_t mapoff = (size_t)delta_data_offset;
1188 				err = got_inflate_to_mem_mmap(&base_buf,
1189 				    &base_bufsz, pack->map, mapoff,
1190 				    pack->filesize - mapoff);
1191 			} else {
1192 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1193 				    == -1) {
1194 					err = got_error_from_errno();
1195 					goto done;
1196 				}
1197 				err = got_inflate_to_mem_fd(&base_buf,
1198 				    &base_bufsz, pack->fd);
1199 			}
1200 			if (err)
1201 				goto done;
1202 			n++;
1203 			continue;
1204 		}
1205 
1206 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1207 		    delta->delta_buf, delta->delta_len, accum_buf,
1208 		    &accum_size, max_size);
1209 		n++;
1210 		if (err)
1211 			goto done;
1212 
1213 		if (n < deltas->nentries) {
1214 			/* Accumulated delta becomes the new base. */
1215 			uint8_t *tmp = accum_buf;
1216 			/*
1217 			 * Base buffer switches roles with accumulation buffer.
1218 			 * Ensure it can hold the largest result in the delta
1219 			 * chain. Initial allocation might have been smaller.
1220 			 */
1221 			if (base_bufsz < max_size) {
1222 				uint8_t *p;
1223 				p = reallocarray(base_buf, 1, max_size);
1224 				if (p == NULL) {
1225 					err = got_error_from_errno();
1226 					goto done;
1227 				}
1228 				base_buf = p;
1229 				base_bufsz = max_size;
1230 			}
1231 			accum_buf = base_buf;
1232 			base_buf = tmp;
1233 		}
1234 	}
1235 
1236 done:
1237 	free(base_buf);
1238 	if (err) {
1239 		free(accum_buf);
1240 		*outbuf = NULL;
1241 		*outlen = 0;
1242 	} else {
1243 		*outbuf = accum_buf;
1244 		*outlen = accum_size;
1245 	}
1246 	return err;
1247 }
1248 
1249 const struct got_error *
1250 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1251     FILE *outfile, FILE *base_file, FILE *accum_file)
1252 {
1253 	const struct got_error *err = NULL;
1254 
1255 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1256 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1257 
1258 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1259 		if (obj->pack_offset >= pack->filesize)
1260 			return got_error(GOT_ERR_PACK_OFFSET);
1261 
1262 		if (pack->map) {
1263 			size_t mapoff = (size_t)obj->pack_offset;
1264 			err = got_inflate_to_file_mmap(&obj->size, pack->map,
1265 			    mapoff, pack->filesize - mapoff, outfile);
1266 		} else {
1267 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1268 				return got_error_from_errno();
1269 			err = got_inflate_to_file_fd(&obj->size, pack->fd,
1270 			    outfile);
1271 		}
1272 	} else
1273 		err = dump_delta_chain_to_file(&obj->size, &obj->deltas, pack,
1274 		    outfile, base_file, accum_file);
1275 
1276 	return err;
1277 }
1278 
1279 const struct got_error *
1280 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1281     struct got_object *obj, struct got_pack *pack)
1282 {
1283 	const struct got_error *err = NULL;
1284 
1285 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1286 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1287 
1288 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1289 		if (obj->pack_offset >= pack->filesize)
1290 			return got_error(GOT_ERR_PACK_OFFSET);
1291 		if (pack->map) {
1292 			size_t mapoff = (size_t)obj->pack_offset;
1293 			err = got_inflate_to_mem_mmap(buf, len, pack->map,
1294 			    mapoff, pack->filesize - mapoff);
1295 		} else {
1296 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1297 				return got_error_from_errno();
1298 			err = got_inflate_to_mem_fd(buf, len, pack->fd);
1299 		}
1300 	} else
1301 		err = dump_delta_chain_to_mem(buf, len, &obj->deltas, pack);
1302 
1303 	return err;
1304 }